New Graduate Student Orientation

Wednesday, January 14, 2009,
5:30pm to 8:30pm,
Enterprise hall room 80

Abstract

For the fourth year, the Volgenau School of Information Technology and Engineering is welcoming its newly admitted graduate students to a special orientation event.

Although orientation is not mandatory, it is highly recommended that both domestic and international students plan to attend. Essential information regarding university services for graduate students and program information from academic departments will be provided. Also, this is your opportunity to meet your peers, the administrative and academic staff members who will assist you during the pursuit of your graduate course work and degree.

Students admitted for the Spring 2009 semester:

Date: Wednesday, January 14, 2009 Time: 5:30-8PM Location: Fairfax Campus, Enterprise Hall, Room 80

The following departments and majors will be represented:

* Applied Information Technology * Civil, Environmental & Infrastructure Engineering * Computer Science * E-Commerce * Electrical & Computer Engineering * Information Security & Assurance * Information Systems * Information Technology * Real Estate Development * Software Engineering * Statistics * Systems Engineering & Operations Research * Telecommunications

RSVP Please RSVP online for orientation by visiting: http://ite.gmu.edu/graduateresearch/responseform/

INFS 501 and INFS 515 Foundation Test-out Exams

Thursday, January 15, 2009,
2:00pm and 3:30pm,
Science and Tecnology II Room 9

Abstract

January 15, Science & Technology II Room 9 INFS 501: 2:00 PM INFS 515: 3:30 PM

Exam Preparation In preparation for an examination, you may wish to review the following textbooks. INFS 501: Discrete Math

Discrete Mathematics with Applications, third edition, by Susanna S. Epp, Published by Brooks Cole, ISBN: 0-534-94446-9.

Syllabus o Chapter 1: The Logic of Compound Statements: 1.1, 1.2, 1.3 o Chapter 2: The Logic of Quantified Statements: 2.1, 2.2, 2.3, 2.4 o Chapter 3: Elementary Number Theory and Methods of Proof: 3.1, 3.2, 3.3, 3.4, 3.5, 3.6 o Chapter 4: Sequences and Mathematical Induction: 4.1, 4.2, 4.3 o Chapter 5: Set Theory: 5.1, 5.2, 5.3 o Chapter 7: Functions: 7.1, 7.2, 7.4 o Chapter 8: Recursion: 8.1, 8.4 o Chapter 9: O-Notation and the Efficiency of Algorithms: 9.1, 9.2, 9.3, 9.4 o Chapter 10: Relations: 10.1, 10.2, 10.3, 10.5 o Chapter 11: Graphs and Trees: 11.1, 11.2, 11.4, 11.5 o Chapter 12: 12.2

INFS 515: Computer Organization

The Essentials of Computer Organization and Architecture, 2nd Edition, Linda Null and Julia Lobur, Jones and Barlett Publishers, 2006, ISBN: 0-7637-3769-0.

Syllabus o Chapter 1: Introduction - Historical Overview o Chapter 2: Data Representation o Chapter 3: Boolean Algebra and Digital Logic o Chapter 4: MARIE o Chapter 5: Instruction set Architectures o Chapter 6: Memory Concepts o Chapter 7: Input/Output and Storage Systems o Chapter 8: System Software o Chapter 9: Alternative Architectures o Chapter 10: Embedded Systems o Chapter 11: Performance Measurement and Analysis

INFS 519 and SWE 510 Foundation Test-out Exams

Friday, January 16, 2009,
2:00pm and 3:30pm,
Science and Technology II room 9

Abstract

January 16, Science & Technology II Room 9 INFS 519: 2:00 PM SWE 510: 3:30 PM

Exam Preparation In preparation for an examination, you may wish to review the following textbooks.

INFS 519: Program Design and Data Structures

Data Structures and Problem Solving Using JAVA, by Weiss, published by Addison-Wesley, ISBN: 0-201-74835-5.

Syllabus o Chapter 1: Primitive Java o Chapter 2: References o Chapter 3: Objects and Classes o Chapter 4: Inheritance o Chapter 5: Algorithm Analysis o Chapter 6: Collections API o Chapter 7: Recursion o Chapter 8: Sorting Algorithms o Chapter 11: Stacks and Compilers o Chapter 14: Graphs o Chapter 16: Stacks and Queues o Chapter 17: Linked Lists o Chapter 18: Trees o Chapter 19: Binary Search Trees o Chapter 20: Hash Tables o Chapter 21: Priority Queue: The Binary Heap

SWE 510: Object-Oriented Programming in Java

Java Gently, by Judith Bishop, 3rd Edition, Addison-Wesley, 2001, ISBN: 0-201-71050-1.

Syllabus o Chapter 1: Introduction o Chapter 2: Basic of Object Orientation o Chapter 4: Input and Output o Chapter 5: Flow of Control o Chapter 6: Arrays and Tablees o Chapter 9: Abstraction and Inheritance o Chapter 10: Graphics and User Interfaces o Chapter 11: Event-driven Programming o Chapter 12: Applets and Servlets o Chapter 13: Multithreading

GRAND Seminar: Reciprocal Velocity Obstacles for Real-Time Multi-Agent Navigation

Tuesday, January 27, 2009,
2:00pm,
ST2, 430
Jur van den Berg

Postdoctoral Researcher
Department of Computer Science
University of North Carolina, Chapel Hill

Abstract

We propose a new concept ---the "Reciprocal Velocity Obstacle"--- for real-time multi-agent navigation. We consider the case in which each agent navigates independently without explicit communication with other agents. Our formulation is an extension of the Velocity Obstacle concept, which was introduced for navigation among (passively) moving obstacles. Our approach takes into account the reactive behavior of the other agents by implicitly assuming that the other agents make a similar collision-avoidance reasoning. We show that this method guarantees safe and oscillation-free motions for each of the agents. We apply our concept to navigation of thousands of agents in densely populated environments containing both static and moving obstacles, and we show that real-time and scalable performance is achieved in such challenging scenarios.

Short Bio

MSc: University of Groningen, The Netherlands PhD: University of Utrecht, The Netherlands. Advisor: prof. Mark Overmars currently: Postdoctoral Researcher at University of North Carolina with profs. Dinesh Manocha and Ming Lin next year: Postdoctoral Researcher at UC Berkeley with prof. Ken Goldberg

Software Engineering Seminar: GMU Software Engineering Seminar Series

Wednesday, January 28, 2009,
12:30 - 01:30PM,
430A ST2
Wei Ding

Assistant Professor, University of Massachusetts Boston

Abstract

Numerous properties of an increasing number of sites on Earth are now thoroughly measured due to routinization of remote sensing and the progress in the GIS techniques. This new found richness of geospatial data in conjuncture with intelligent data processing transforms our understanding of geospatial phenomena. This project seeks to contribute to this trend by developing methods for auto-modeling causal relationships that are present between geographically collocated geo-variables but have been difficult or impossible to identify using traditional methods. Specifically, the project aims at providing the scientific community with an efficient tool for auto-analyzing the root causes behind observed patterns in geospatial data. We propose to integrate techniques from association analysis, reinforcement learning, and similarity measurement to design a tool that provides a domain expert with a complete yet comprehensible empirical model of an observed phenomenon. Novel solutions are required at all stages of the method development: (i) integration of geospatial data into the format suitable to association analysis, (ii) determining a scope and footprint of the studied phenomenon, (iii) selecting patterns of controlling factors that are indicative of the phenomenon, and (iv) summarizing the patterns into the products which are of immediate use to a domain experts.

Speaker's Bio

Wei Ding is an Assistant Professor of Computer Science at the University of Massachusetts Boston. She received her M.Sc. Degree in Software Engineering from the George Mason University in 2000 and her Ph.D. degree in Computer Science from the University of Houston in 2008. Her main research interests include Data Mining, Machine Learning, Artificial Intelligence and with applications to astronomy, geosciences, and environmental sciences.

SANG Seminar: Sampling Attacks Against Hidden Web Databases

Friday, January 30, 2009,
03:00PM,
ST2, Room 430A
Prof. Nan Zhang

Abstract

A large number of online databases are hidden behind form-like interfaces which allow users to execute search queries by specifying selection conditions in the interface. Most of these interfaces return restricted answers (e.g., only top-k of the selected tuples), while many of them also accompany each answer with the COUNT of the selected tuples. We shall present techniques which leverage the COUNT information to efficiently acquire unbiased samples of the hidden database. We also discuss variants for interfaces which do not provide COUNT information.

* Collaborative work with Arjun Dasgupta and Dr. Gautam Das of the University of Texas at Arlington

Bio

Dr. Nan Zhang is an Assistant Professor of Computer Science at the George Washington University. Prior to joining GWU, he was an assistant professor of Computer Science and Engineering at the University of Texas at Arlington from 2006 to 2008. He received the B.S. degree from Peking University in 2001 and the Ph.D. degree from Texas A&M University in 2006, both in computer science. His current research interests span security and privacy issues in databases, data mining, and computer networks, including privacy and anonymity in data collection, publishing, and sharing, privacy-preserving data mining, and wireless network security and privacy. He received the NSF CAREER award in 2008.

GRAND Seminar: A Survey of Approximate Convex Hull Algorithms

Tuesday, Feburary 10, 2009,
12:00pm,
ST2 430
Raheem Rufai

Department of Computer Science
George Mason University

Abstract

The convex hull of a point set S, often denoted as conv(S), is the smallest convex set that contains S. This essentially means that a line connecting any pair of vertices in S is enclosed by conv(S). This talk will introduce the convex hull problem and briefly survey existing state of the art for exact algorithms. The crux of the discussion, however, will focus on the current state of affairs of approximation algorithms for the convex hull. The talk might also touch upon related issues such as kinetization and dynamization of these algorithms and the core-set framework as time permits.

Short Bio

Raimi Rufai is a doctoral student in the CS Department, George Mason University. His research interests span Computational Geometry, Software Engineering and enterprise software development and modeling. He is currently working on his dissertation in Computational Geometry under the tutelage of Prof. Dana Richards.

SANG Seminar: Preventing SQL Injection Attacks: A Policy-based Data Type Checking

Friday, February 13, 2009,
03:00PM,
ST2 430A
Anyi Liu

Abstract

SQL injection attacks (SQLIAs) use maliciously crafted SQL input to force web application to function differently from what the query designed to be. Several researchers have developed intrusion prevention techniques based on input validation, learning, or static analysis. A weakness of these techniques is that they primarily focused on identifying malicious input, but not the output. To address this weakness, we develop a policy-based type checking approach that first identifies all the database access points, then use security policy to enforce each of them. The novelty of our approach, as compare to other SQLIAs prevention techniques, is that it focuses on the meaning of the query and the necessary information sending between the program and the database. An interesting aspect of our technique is that when the policies integrate with the control flow graph of user of different privilege level, it can significantly increase the precision of prevention. To evaluate the effectiveness and the performance of our system, we implemented a prototype system which tested by real SQL attacks. We demonstrate that our approach is robust enough and holds promise for more precise detection.

Bio

Anyi Liu is a Ph.D. student in the Department of Information and Software Engineering/Computer Science at George Mason University. He received a BS and MS in Computer Science from Dalian University of Technology, China, in 1997 and 2001, respectively. He is now pursuing a Ph.D. in Information Technologies from George Mason University. His research interests include information security, intrusion detection/prevention, and security issues of web applications.

CS/GRAND Seminar: Lifelines2: Interactive Visualization of Temporal Categorical Data

Tuesday, February 17, 2009,
2:00PM,
ST2, 430
Taowei David Wang

Ph.D candidate
Department of Computer Science
University of Maryland, College Park

Abstract

As large data repositories of temporal data become more prevalent today, it also becomes more important to develop appropriate techniques to support visual exploratory tasks on this type of data. Although much work has been done to develop techniques for numerical time series, temporal categorical data has been mostly overlooked. This type of data is best characterized by events that have names and time stamps. To be able to study the ordering of events and the prevalence of ordering can reveal interesting relationships among the different events, and consequently help analysts formulate new hypotheses, gain new insights. I present Lifelines2, an interactive visualization system designed for visual analysis of temporal categorical data across multiple records.

At the center of Lifelines2 is the Align-Rank-Filter (ARF) framework. It allows users to quickly manipulate how they see the data by their features. Alignment lets users specify a reference event, and subsequently change the view of data to be relative to that event. This facilitates the discovery of important temporal relationships associated with the reference events. Rank and Filter are old ideas that let users reorder and trim the data. Coupled with alignment, however, they also offer new affordances to support more sophisticated exploratory tasks.

In this talk, I present the features of Lifelines2, and show a video on how a medical researcher might use Lifelines2 to identify patients who exhibit a certain pattern of medical events. I will also present the results from our controlled study, which shows that the alignment feature is instrumental in aiding users to perform complex tasks. Finally, I will talk about the first impressions of Lifelines2 from our collaborators, and peek at our future research on the topic.

Short Bio

Taowei David Wang is a Ph.D candidate in the Department of Computer Science in the University of Maryland. He is part of the Human Computer Interaction Lab, and is advised by Ben Shneiderman. His dissertation topic is on interactive visualization techniques of temporal categorical data for the purpose of hypothesis generation. He is currently working closely with physicians and clinicians on searches over electronic health records for this purpose, and is scheduled to graduate in the winter of 2009.

SANG Seminar: One Cell is Enough to Break Tor's Anonymity

Friday, February 20, 2009,
02:00 - 03:30 pm,
ST2, Room 430A
Prof. Xinwen Fu

Abstract

Tor is a real-world, circuit-based low-latency anonymous communication network, supporting TCP applications over the Internet. In this talk, we will present a new class of attacks, protocol-level attacks, against Tor. Different from existing attacks, these attacks can confirm anonymous communication relationships quickly and accurately by manipulating one single cell and pose a serious threat against Tor. In protocol-level attacks, a malicious entry onion router may duplicate, modify, insert, or delete cells of a TCP stream from a sender. The manipulated cells traverse middle onion routers and arrive at an exit onion router along a circuit. Because Tor uses the counter mode AES (AES-CTR) for encrypting cells, the manipulated cells disrupt the normal counter at exit onion routers and decryption at the exit onion router incurs cell recognition errors, which are unique to the investigated protocol-level attacks. If an accomplice of the attacker at the entry onion router also controls the exit onion router and recognizes such cell recognition errors, the communication relationship between the sender and receiver will be confirmed. Protocol-level attacks can also be used for launching the denial-of-service (DoS) attack to disrupt the operation of Tor. We systematically analyze the impact of these attacks. We have implemented these attacks on Tor and our experiments validate their effectiveness and efficiency. We also present guidelines for defending against such attacks.

Speaker's bio

Dr. Xinwen Fu is an assistant professor in the Department of Computer Science, University of Massachusetts Lowell. He received his B.S. (1995) and M.S. (1998) in Electrical Engineering from Xi'an Jiaotong University, China and University of Science and Technology of China respectively. He obtained his Ph.D. (2005) in Computer Engineering from Texas A&M University. Dr. Fu's current research interests are in network security and privacy, information assurance, computer forensics, system reliability and networking QoS. He has been publishing papers in conferences such as IEEE Symposium on Security and Privacy (S&P), IEEE International Conference on Computer Communications (INFOCOM) and IEEE International Conference on Distributed Computing Systems (ICDCS), journals such as IEEE Transactions on Parallel and Distributed Systems (TPDS), and book chapters. His research is supported by NSF.

Faculty Candidate Seminar: Highly Interactive Scalable Virtual Worlds

Friday, February 27, 2009,
11:00 am,
JC Gold Room
Dr. Graham Morgan

Newcastle University

Abstract

The arrival, in the past decade, of commercially successful virtual worlds used for online gaming and social interaction has emphasized the need for a concerted research effort in this media. A pressing problem is that of incorporating ever more elaborate gaming scenarios into virtual worlds while ensuring player numbers can be measured in the millions. In this talk I will describe the work I have led over the past five years in this area. The development of a highly scalable platform for online gaming will be described and a number of performance results presented

Biography

Graham Morgan is a lecturer in the School of Computing Science. He received his PhD in Computing from Newcastle University in 2000. He was appointed to a Lecturer position in September 2000 and is a member of the Distributed Systems Research Group. He is currently a visiting faculty member with the Department of Computer Science at George Mason University.

Dr. Morgan’s interests are in : game engineer technologies, online gaming, physics simulations, fault-tolerant computing, scalable server-side clustered solutions, collision detection, group communications, distributed transactions.

SANG/GRAND Seminar: Online Scheduling of Packets with Deadlines in a Bounded Buffer

Tuesday, March 03, 2009,
12:00PM,
ST2 430
Fei Li

Assistant Professor
Department of Computer Science
George Mason University

Abstract

Motivated by the Quality-of-Service buffer management problem, we consider online scheduling of packets with deadlines in a size-bounded buffer. This model is called bounded-buffer model. Time is discrete. Packets arrive at a buffer over time. Each packet p has an integer release time r_p, a non-negative value w_p, and an integer deadline d_p. d_p specifies the time by which p should be sent. If p is transmitted by its deadline d_p, p contributes our objective its value w_p. At any time, the buffer can store no more than b packets. In each time step, at most one packet can be sent. This model is preemptive: packets already existing in the buffers can be dropped at any time before they are sent and the dropped packet cannot be delivered any more. Our objective is to maximize the total value of the packets sent by their deadlines. This bounded-buffer model generalizes the bounded-delay model proposed in (INFOCOM 00, CISS 01, STOC 01) in which the buffer's size b is assumed larger than any packet's slack time (a packet p's slack time is defined as d_p - r_p).

In SWAT 06, Azar and Levy consider a model called multi-buffer model with multiple size-bounded buffers. The lower bound of competitive ratio 1.618 (INFOCOM 00, CISS 01, Algorithmica 03) for the bounded-delay model applies to the multi-buffer model. The authors (Azar and Levy, SWAT 2006) developed a deterministic 9.82-competitive algorithm which applies to the bounded-buffer model as well.

For the bounded-buffer model, we present a deterministic 3-competitive online algorithm and a randomized 2.618-competitive online algorithm. We also show that the lower bound of competitive ratio for a broad family of deterministic algorithms is improved from 1.618 to 2.

Short Bio

Dr. Fei Li is an assistant professor at Computer Science Department of George Mason University since Fall 2007. His major research interests are online algorithms, approximation algorithms, randomized algorithms and online learning algorithms. He got his PhD in February 2008 from Columbia University, in computer science.

Faculty Candidate Seminar: Creating 3D Animated Human Behaviors for Virtual Worlds

Friday, March 06, 2009,
11:00 am,
Research I 163
Jan Allbeck

University of Pennsylvania

Abstract

As we journey through our day, our lives intersect with other people. We see people leaving for work, waiting for trains, meeting with friends, hard at work, and thousands of other activities that we may not even be conscious of. People create a rich tapestry of activity throughout our day, a "human texture". We may not always be aware of this texture, but we would definitely notice if it were missing, and it is missing from many simulations and games. Most crowd simulations to date include only basic locomotive behaviors possibly coupled with a few stochastic actions or are meticulously hand scripted. I will describe an architecture, called CAROSA (Crowds with Aleatoric, Reactive, Opportunistic, and Scheduled Actions), that facilitates the creation of heterogeneous populations for simulations by using a commercial off-the-shelf software package (Microsoft Outlook®), a Parameterized Action Representation (PAR), and crowd simulation software (HiDAC). We incorporate four different broad action types into CAROSA: scheduled, reactive, opportunistic, and aleatoric. Scheduled activities arise from specified roles for individuals or groups; reactive actions are triggered by contextual events or environmental constraints; opportunistic actions arise from explicit goals and priorities; aleatoric actions are random but structured by choices, distributions, or parametric variations. The CAROSA architecture enables the specification and control of actions for more realistic “human textures” in virtual worlds such as buildings and cities, links human characteristics and high level behaviors to animated graphical depictions, and relieves some of the burden in creating and animating heterogeneous 3D animated populations.

Biography

Jan Allbeck is a Ph.D. candidate in the Department of Computer and Information Science at the University of Pennsylvania. She is also Associate Director of the Center for Human Modeling and Simulation, where she coordinates and participates in the research projects affiliated with HMS as well as coordinating the operational aspects of the lab facility, the SIG Center for Computer Graphics. HMS is a part of CG@Penn which also includes the Digital Media Design undergraduate program and a masters program, Computer Graphics and Game Technology. She has a Bachelors degrees in Mathematics and Computer Science from Bloomsburg University and a Masters degree in Computer Science from Penn. She has had the great opportunity to explore many aspects of computer graphics, but is most drawn to research at the crossroads of animation, artificial intelligence, and psychology in the simulation of virtual humans. Her current research focuses on the creation and simulation of functional crowds.

Faculty Candidate Seminar: Heuristics are important for improving the performance of search-based algorithms

Monday, March 16, 2009,
11:00 am,
JC Gold Room
Dr. Nathan Sturtevant

University of Alberta

Abstract

Pattern databases are the most common form of memory-based heuristics, and have been well-studied over the last decade. But, in many domains pattern databases are ineffective at improving heuristic estimates. In this talk I will describe several of these domains, including path-finding for commercial video games, and motivate how improved heuristics can be used. I will then present new research on canonical and differential heuristics. These heuristics provide an order of magnitude or larger reduction in search effort over the previous best-known techniques.

Biography

Nathan Sturtevant is currently an adjunct professor in Computer Science at the University of Alberta, in Edmonton, Canada. He received his bachelor's degree from UC Berkeley and his PhD in from UCLA in 2003. His main research focus is in heuristic search with an interest in single- and multi-player games. Nathan's techniques have been implemented in BioWare's upcoming game, Dragon Age (http://dragonage.bioware.com/). Nathan is also known as the author of the popular 1990s Mac shareware game, Dome Wars.

Faculty Candidate Seminar: Analysis and Control for Biological Networks

Thursday, March 19, 2009,
11:00am,
SUB II Room 5/6
Dr. Xiaoning Qian

Texas A & M University

Abstract

The ultimate goal of studying the systems biology of biological networks is to apply intervention to living organisms for effective treatment of diseases. Starting with a project of comparing protein interaction networks of different organisms, I will present an efficient framework based on hidden Markov models (HMMs) to identify conserved homologous pathways in networks of interest. Finding these common interaction patterns across or within organisms can lead to a better understanding of the regulatory mechanisms underlying various cellular functions. With the regulatory relationship learned either from prior knowledge or interaction measurements, we further analyze long-run genome behavior based on its underlying Markov chain in the framework of the probabilistic Boolean network (PBN) model. The change of steady-state distribution of the Markov chain caused by possible perturbations to a network is the key measure for intervention. We derive analytic results for changes in the steady-state distributions of probabilistic Boolean networks resulting from modifications to the underlying regulatory rules. From these analytic results, we derive intervention strategies to obtain therapeutic benefits for future drug design or gene therapy design. The preliminary results in a network modeling melanoma cell line have shown that our methods can potentially serve as future intervention strategies to identify potential drug targets and design gene-based therapeutic strategies.

Biography

Xiaoning Qian obtained his Bachelor and Master degrees in Electronic Engineering from Shanghai Jiao-Tong University in China in 1997 and 1999. After receiving his Ph.D. degree in Electrical Engineering for his work on shape-based indexing in medical image databases at Yale University in 2005, he worked as a postdoctoral associate in the Department of Diagnostic Radiology at Yale on several biomedical image analysis projects. In 2007, he joined the Bioinformatics Training Program at Texas A&M University as an associate research scientist in the Department of Electrical and Computer Engineering and a research assistant professor in the Department of Statistics. His current main research focus is to develop statistical and geometric models and relevant algorithms for the inference, analysis, and intervention of biological

Faculty Candidate Seminar: The Role of Computation in Cellular and Molecular Investigations of Human Disease

Tuesday, March 24, 2009,
10:00am,
SUB II room 1/2
Austin Huang

Harvard/MIT

Abstract

Computational approaches can play an important role in improving our understanding of diseases. I will discuss two critical applications of computation – the use of image processing and computer vision algorithms for quantitative analysis of bioimaging data and molecular modeling of disease-associated proteins. Two examples from my research will be discussed - image analysis of airway epithelial cells under conditions mimicking an asthma attack and computational modeling of conformational preferences of tau protein in Alzheimer’s disease. The presentation is for a general audience – extensive background in computational biology is not assumed.

Research Seminar: Efficient Algorithms for Protein Structure-Sequence Alignment and Applications

Thursday, March 26, 2009,
10:00am,
Research I 163
Dr. Xiuzhen Huang

Arkansas State University

Abstract

The immediate need for knowing tertiary proteins structure intensifies as the applications of engineered proteins for drugs, carriers, enzymatic activities, receptors, vaccines, antibodies, biomaterials and nanotechnology, in vitro synthesis, and detection systems grow. Due to the high cost of experimental determination of protein structures, there is tremendous demand for efficient and accurate computational predictions of protein structures from primary sequence data and inference of protein functions from structure for rational protein or drug design. However, current computational approaches provide limited accuracy, often cannot handle large proteins, and require significant computational time. We developed a novel parameterized algorithm for structure-sequence alignment, which is the highly challenging computational part for protein structure prediction based on threading. The algorithm incorporates a graph theoretic model, parameterized computation and tree decomposition techniques. Experimental results demonstrate a successful application of the algorithm to non-coding RNA structure search in genomes. Methods based on this algorithm provide analogous increases in efficiency and accuracy in response to the protein structure prediction challenges. Through collaborative research with biologists and biomedical researchers, we are applying these methods to structure-function studies on proteins (like RTB and UGTs) with direct significant medical applications.

Biography

Xiuzhen Huang received the PhD degree in Computer Science from Texas A&M University in 2004. Since then she has been working as an assistant professor in the Department of Computer Science, Arkansas State University. She is currently the principal investigator of the bioinformatics lab at Arkansas Bioscience Institute (ABI). She holds an adjunct appointment in Information Science Department, University of Arkansas at Little Rock (UALR), is the graduate faculty member of the Molecular Biosciences PhD Program at ABI, and the joint PhD Program in Bioinformatics of UALR and University of Arkansas for Medical Sciences (UAMS). Her research strengthens and foci are in bioinformatics and computational biology, algorithm design and analysis. Her work includes collaborative projects with biologists at ABI and UALR, and biomedical researchers at UAMS, and has been supported by NIH INBRE and NSF EPSCoR grants for which she serves as the PI or a co-PI.

Sotware Engineering Seminar: An Experimental Comparison of Four Unit Test Criteria: Mutation, Edge-Pair, All-uses and Prime Path Coverage

Wednesday, March 28, 2009,
12:00pm - 01:00pm,
Room 430, SCII
Nan Li

Abstract

In this talk, I present the results from a comparison of four unit level software testing criteria. Mutation testing, prime path coverage, edge-pair coverage, and all-uses testing were compared on two bases: the number of seeded faults found and the number of tests needed to satisfy the criteria. The comparison used Java classes and hand-seeded faults. Tests were designed and generated mostly by hand with help from tools that compute test requirements and mu-Java. I also present a secondary measure, a cost benefit ratio, computed as the number of tests needed to detect each fault. I also discuss some specific faults that were not found and present analysis for why not.

Bio

Nan Li is a PhD student in Computer Science Department, Volgenau School of Information Technology and Engineering. His current research mainly focuses on software testing and he is also interested in other fields of software engineering.

Sotware Engineering Seminar: Comparison of Unit-Level Automated Test Generation Tools

Wednesday, March 28, 2009,
12:00pm - 01:00pm,
Room 430, SCII
Shuang Wang

Abstract

Data from projects worldwide show that many software projects fail and most are completed late or over budget. Unit testing is a simple but effective technique to improve software in aspects of quality, flexibility, and time to market. However, testing each unit by hand is very expensive, possibly prohibitively so. Automation is essential to support unit testing and as unit testing is achieving more attention, developers are using automated unit testing tools more often. However, developers have very little information about which tools are effective. This experiment compares three well-known public-accessible unit test tools, JCrasher, TestGen4j, and JUB. We apply these tools to a variety of Java programs and evaluate them based on their mutation scores. As a comparison, we manually created two additional sets of tests. One test set contained random values with the same number of tests the three test tools created, and the other contained values to satisfy edge coverage.

Bio

Shuang Wang is a PhD student and a teaching assistant in Computer Science Department, Volgenau School of Information Technology and Engineering, George Mason University. Her current interests include software testing, Web Application testing, and Mutation testing. Her advisor is Dr. Jeff Offutt.

GRAND Seminar: Towards a Universal Text Classifier: Transfer Learning from Encyclopedic Knowledge

Tuesday, March 31, 2009,
12:00PM,
ST2 430
Pu Wang

Ph.D Student, CS, GMU

Abstract

Document classification is a key task for many text mining applications. However, traditional text classification requires labeled data to construct reliable and accurate classifiers. Unfortunately, labeled data are seldom available, and often too expensive to obtain. In this work, we propose a universal text classifier, which does not require any labeled training document. Our approach simulates the capability of people to classify documents based on background knowledge. As such, we build a classifier that can effectively group documents based on their content, under the guidance of few words describing the classes of interest. Background knowledge is modeled using encyclopedic knowledge, namely Wikipedia. Wikipedia's articles related to the specific problem domain at hand are selected, and used during the learning process for predicting labels of test documents. The universal text classifier can also be used to perform document retrieval, in which the pool of test documents may or may not be relevant to the topics of interest for the user. In our experiments with real data we test the feasibility of our approach for both the classification and retrieval tasks. The results demonstrate the advantage of incorporating background knowledge through Wikipedia, and the effectiveness of modeling such knowledge via probabilistic topic modeling. The accuracy achieved by the universal text classifier is comparable to that of a supervised learning technique for transfer learning.

Short Bio

Pu Wang is a PhD student in the Department of Computer Science at George Mason University. He received a Masters degree in Computer Science from Beijing University in 2007, and was an intern at Microsoft Research Asia from May 2006 to July 2007. His research interest is in machine learning, currently focusing on graph learning.

SANG Seminar: Defeating Anti-detection for Application-level Malware Analysis**

Friday, April 10, 2009,
3:00-4:30PM,
ST2 430A
Lei Liu

Abstract

Malware analysis is critical for malware detection and prevention. To defeat malware analysis and detection, today malware commonly adopts various sophisticated anti-detection techniques. For example, malware often performs various debugger, emulator, and virtual machine fingerprinting. These mechanisms produce more and more stealthy malware that challenges malware analysis schemes.

In this work, we propose Malyzer to defeat malware anti-detection mechanisms at startup and runtime so that malware behavior during execution can be accurately captured and distinguished. For analysis, Malyzer starts a process copy, referred to as a shadow process, on the same host by defeating anti-detection mechanisms. Since ultimately malware will conduct local information harvesting or dispersion, Malyzer constantly monitors the shadow process's behavior and adopts a hybrid scheme for its behavior analysis. In our experiments, Malyzer can accurately detect all malware samples that employ various anti-detection techniques.

Bio

Lei Liu is a Ph.D. student in Computer Science Department of George Mason Univesity. His research interests incude system security, network application, operation systems and sogrware engineering. Before pursuing his Ph.D. degree, he has several years' experience in IT industry.

GRAND Seminar: Affine Invariant-Based Classification of Inliers and Outliers for Image Matching

Tuesday, April 21, 2009,
12:30PM,
Engineering Building Room 4201
Dan Fleck

Department of Computer Science
GMU

Abstract

This presentation describes a new approach to classify tentative feature matches as inliers or outliers during wide baseline image matching. Wide-baseline matching is the process of matching one image to another. After typical feature matching algorithms are run and tentative matches are created, our approach is used to classify matches as inliers or outliers to a transformation model. The approach uses the affine invariant property that ratios of areas of shapes are constant under an affine transformation. Thus, by randomly sampling corresponding shapes in the image pair we can generate a histogram of ratios of areas. The matches that contribute to the maximum histogram value are then candidate inliers. The candidate inliers are then filtered to remove any with a frequency below the noise level in the histogram. The resulting set of inliers are used to generate a very accurate transformation model between the images. In our experiments we show similar accuracy to the standard RANSAC approach and an order of magnitude efficiency increase using this affine invariant-based approach.

Short Bio

Dan Fleck is currently an instructor of Computer Science at George Mason University (GMU). He earned his B.S. in Electrical Engineering from the University of Texas at Austin before moving to Northern Virginia. While working full-time Dan completed his M.S. in Software Engineering from GMU.

Currently he is pursuing a doctorate degree in Computer Science. Dan's doctoral research is in Computer Vision under Dr. Zoran Duric. Specifically researching novel approaches to matching images taken at different viewing angles, locations and scales.

Previously, Dan was a technical lead and project manager at SRA International. At SRA he led projects ranging from 5 to 50 people for a variety of government clients. Dan worked within SRA's Health Systems group, Data Mining Center and most recently as technical lead within the Advanced Technology Group. Dan continues to serve in an advisory role at SRA.

PhD Dissertation Defense: Routing in Delay Tolerant Networks

Thursday, April 23, 2009,
01:00pm,
Suite 2500 in the Engineering Building
Muhammad Abdulla

PhD Student, Department of Computer Science

Abstract

Delay tolerant networks (DTNs) are a newly emerging class of mobile distributed systems that are designed for extreme environments such as disaster response, wildlife management and planned networks in space. DTNs require a re-design of existing mobile network routing protocols because fully connected end-to-end paths between senders and receivers may never exist. The design and analysis of efficient DTN routing protocols is largely based upon a fundamental understanding of the mobility characteristics of individual nodes. My research focuses on providing performance models for DTN node mobility as well as basic routing protocols for unicast and multicast applications. I provide analytical results for parameters such as nodal contact and inter-contact times, and propose and evaluate several novel modeling techniques for evaluating multicopy DTN routing. Using the insights from these techniques I have designed and tested two DTN routing protocols. The first is a core-based technique for unicast routing, and the second is CERM, Controlled Epidemic Routing for Multicasting. Experimental results demonstrate that my approach has a highly accurate predictive value for routing performance and protocol efficiency, and can be used in a wide-variety of application settings.

SANG Seminar: Open Problems in Vehicular Ad Hoc Network Security

Friday, April 24, 2009,
3:00 - 4:30PM,
New Engineering Building, Room 4201
Eric John Swankoski

Abstract

In the past decade, more and more research has been done on nearly every aspect of vehicular networks, which are quite likely to become popular on the world's highways. Perhaps not surprisingly, security has become an issue. As with any network, attacks may come from a variety of directions. Vehicular networks present unique challenges in that the network itself is markedly atypical in terms of topology and node mobility. In addition, unmitigated attacks carry potentially very significant consequences. High node mobility coupled with frequent network changes makes traditional authentication and security mechanisms impractical in many cases. The problem becomes then the authentication of the data content within the network given the presence of adversaries. We identify probable attacks against vehicle-to-vehicle message applications and propose potential mitigation methods.

Bio

Eric John Swankoski is a third-year Ph.D. student in Dept. of Computer Science at George Mason University. He obtained his B.S. and M.S. at Penn State in 2002 and 2004, respectively.

SWE Seminar: Game Theory for Dummies

Thursday, April 30, 2009,
12:00pm - 01:00pm,
Engineering Building, Room 3507
Zeynep Zengin

PhD student in the Department of Computer Science

Abstract

Optimization problems are computational problems that address to find the best solution among all feasible solutions. This field of research sits in the intersection of Artificial Intelligence, Mathematics, and Software Engineering. An optimization can be the interest of one or more users. One user optimization problems can be like best choice of essentials, shortest traveling route etc. Multi user optimization problems can be network load distribution, bidding optimization, or QoS optimization when the system is used by more than one person. This type of problems requires balancing the outcome within all users. Game theory formalizes the outcomes of each user considering every possible way of action for each user. It formulates the best choice of a user depending of choices of other users. When there is a global or local optimal solution where all users want to stay with their strategies, this situation is called Equilibrium. The desired solution of a multi user optimization problem is to formulate it game theoretically and reach to an equilibrium point.

Bio

Zeynep Zengin is a PhD student in Computer Science Department, Volgenau School of Information Technology and Engineering. She got her bachelor’s degrees on Information Systems with double major of Computer Engineering from Isik University, Istanbul, Turkey. She also holds a Master of Science degree in Computer Engineering from Bogazici University, Istanbul, Turkey. Her current research mainly focuses on end-user development, self adapting systems, ubiquitous computing, and Computation Tree Logic.

SWESeminar: Privacy-Enhanced Trust Management

Thursday, April 30, 2009,
12:00pm - 01:00pm,
Engineering Building, Room 3507
Dalal Al-Arayed

Doctoral candidate in the IT program

Abstract

This talk will focus on research being conducted as part of Dalal Al-Arayed's dissertation which aims to produce a conceptual framework that extends current trust models to cover the needs of smart spaces (Trust in Smart Spaces- TISS); and that is generalized to be applicable across multiple problem scenarios (Generalized- TISS). It will cover the employment of privacy policies right at the beginning of the trust formation function to direct the application of the trust model for decision making. This adds a human factor and enables the user to personalize the trust model.

Bio

Dalal Al-Arayed is a doctoral candidate in the IT program. She has her BS in Computer Science with a minor in Business Administration from Coastal Carolina University, and her MS in IT from the University of North Carolina at Charlotte. She is an international student from Bahrain.

PhD Dissertation Defense: Service Composition Framework to Unify Simulation and Optimization in Supply Chains

Thursday, April 30, 2009,
01:00pm,
Room 4201, Engineering Building
Malak Talal Al-Nory

PhD Student, Department of Computer Science

Abstract

In supply chains, decisions have to be made regularly to coordinate production and delivery of goods or services effectively across enterprises, and to respond to constantly changing and increasingly customized market demands. Decision-making for supply chains requires the implementation of simulation and/or optimization techniques. Object-oriented simulation provides many advantages in ease of model development, testing, and extensibility, yet it is mainly based on trial-and-error, and thus, lacks systematic optimization. Mathematical programming (MP), on the other hand, allows supply chain problems to be solved efficiently using solvers with well-established optimization algorithms. However, MP uses relatively low-level mathematical abstraction, and thus, is challenging for individuals who are not Operations Research experts. This research introduces a novel approach to address these challenges. Proposed in this dissertation is a framework and language that unify the simulation and optimization of supply chains, which provides modelers with the advantages of simulation-like model development, testing, and reuse, and at the same time, the efficiency of MP. The proposed Service Composition (SC) model represents a business transaction as a Service; consequently, the supply chain can be modeled as a Service with interrelated composite or atomic sub-services. A Service is a class in Java programming language in which its constructor computes a business objective associated with the service instance. The constructor may involve decision-choice constructs for one or more decision variables and assert statements. While the procedural language provides the “simulation-like” semantics, the optimization semantics is achieved by automatic compilation of the simulation model into a MP model and solving it using an external solver. The proposed extensible library of simulation-like modeling components allows quick construction of complex supply chain simulation models and the development of new composite services. This dissertation also studied modeling supply chain services that involve uncertainty factors. The proposed two-stage recourse stochastic programming extension provides an environment to model services for decision-making with future corrective response. The decision variables are represented by means of a special structure to capture the random parameters. The effectiveness of our proposed framework is assessed with experiments that compare the performance of automatically compiled supply chain models with expertly crafted ones.

SANG Seminar: A Case Study of Traffic Locality in Internet P2P Live Streaming**

Friday, May 01, 2009,
3:00-4:30PM,
Engineering Building, Room 4201
Yao Liu

Yao Liu is a Ph.D. student of Computer Science Department at
George Mason University

Abstract

With the ever-increasing P2P Internet traffic, recently much attention has been paid to the topology mismatch between the P2P overlay and the underlying network due to the large amount of cross-ISP traffic. Mainly focusing on BitTorrent-like file sharing systems, several recent studies have demonstrated how to efficiently bridge the overlay and the underlying network by leveraging the existing infrastructure, such as CDN services or developing new application-ISP interfaces, such as P4P. However, so far the traffic locality in existing P2P li ve streaming systems has not been well studied.

In this work, taking PPLive as an example, we examine traffic locality in Internet P2P streaming systems. Our measurement results on both popular and unpopular channels from various locations show that current PPLive traffic is highly localized at the ISP level. In particular, we find: (1) a PPLive peer mainly obtains peer lists referred by its connected neighbors (rather than tracker servers) and up to 90 percent of listed peers are from the same ISP as the requesting peer; (2) the major portion of the streaming traffic received by a requesting peer (up to 88 percent in popular channels) is served by peers in the same ISP as the requestor; (3) the top 10 percent of the connected peers provide most (about 70 percent) of the requested streaming data and these top peers have smaller RTT to the requesting peer. Our study reveals that without using any topology information or demanding any infrastructure support, PPLive achieves such high ISP level traffic locality spontaneously with its decentralized, latency based, neighbor referral peer selection strategy. These findings provide some new insights for better understanding and optimizing the network- and user-level performance in practical P2P live streaming systems.

GRAND Seminar: An Introduction to iPhone Development

Friday, May 01, 2009,
12:00PM,
ENGR 4201
Karl Majer

Computer Science, George Mason, University

Abstract

The iPhone is easily one of the most enticing platforms for the new developer to try and stake their claim in the gold-rush environment that Apple has fostered on its application store. Unlike the Open Source realm, programming in the iPhone environment introduces its own complexities and Apple imposed restrictions upon the new developer. Drop in and learn about the issues I encountered during a semester of coding on the iPhone platform. I will discuss building an application from the inception to a working application, talk about the various issues faced along the way, and attempt to answer any questions you may have about bootstrapping your own attempts at finding fame and fortune in the iTunes AppStore.

Short Bio

Karl Majer is a part-time undergraduate student in the Computer Science Department, Volgenau School of Information Technology and Engineering. When not attending class, Karl is an IT Professional with over 15 years of industry experience as an architect, systems administrator, and programmer.

SANG Seminar: Towards Optimal Resource Utilization in Heterogeneous P2P

Friday, May 08, 2009,
03:00 - 04:30,
Engineering Building, Room 4201
Dongyu Liu

PhD Candidate in Dept. of Computer Science

Abstract

Though plenty of research has been conducted to improve Internet P2P streaming quality perceived by endusers, little has been known about the upper bounds of achievable performance with available resources so that different designs could compare against. On the other hand, the current practice has shown increasing demand of server capacities in P2P-assisted streaming systems in order to maintain high-quality streaming to end-users. Both research and practice call for a design that can optimally utilize available peer resources. In the paper, we first present a new design, aiming to reveal the best achievable throughput for heterogeneous P2P streaming systems. We measure the performance gaps between various designs and this optimal resource allocation. Through extensive simulations, we find out that several typical existing designs have not fully exploited the potential of system resources. However, the control overhead prohibits the adoption of this optimal approach. Then, we design a hybrid system in trading off the cost of assignment and utilization of resources. This hybrid approach has a proved theoretical bound on efficiency of utilization. Simulation results show that compared with the optimal resource allocation, our proposed hybrid design can achieve near-optimal (up to 90 percent) utilization while only use much less (below 4 percent) control overhead. Our results provide a basis for both server capacity planning in current P2Passisted streaming practice and future protocol designs.

Bio

Dongyu Liu is a Ph.D. candidate in Dept. of Computer Science at George Mason University. His research interests are in the areas of networking and systems, particular on Internet streaming delivery and QoS.

PhD Dissertation Defense: Online Topic Detection, Tracking, and Significance Ranking Using Generative Models

Tuesday, June 09, 2009,
02:00 - 03:30 pm,
Engineering Building, Room 4201
Loulwah Al-Sumait

PhD Candidate in Department of Computer Science

Abstract

Online processing of text streams is an essential task of many legitimate applications. The objective is to deconstruct the documents into semantically coherent threads or topics, analyze the development of the topics over time, and identify newly emerging topics. This must be accomplished by processing only the textual content and publication time of document streams without requiring any metadata, such as hyperlinks or citation data. This dissertation presents an "Online Topic Model " (OLDA), a generative topic model that automatically captures the thematic patterns and identifies emerging topics in text streams and their changes over time. The proposed approach allows the topic modeling framework, specifically the Latent Dirichlet Allocation (LDA) model, to work in an online fashion such that it incrementally builds an up-to-date model (mixture of topics per document and mixture of words per topic) when a new document (or a set of documents) appears based on the Empirical Bayes method. The idea is to incrementally update the current model according to the information inferred from the new stream of data with no need to access previous data. The dynamics of the proposed approach also provide an efficient means to track drifts in topics and detect the emerging topics in real time. This dissertation also extends the proposed online topic model to incorporate semantic history propagated from models that were estimated within a "sliding history window." Since the proposed approach is totally unsupervised and data-driven, the effect of different factors is analyzed, including the window size, history weight, and equal/decaying history contribution. Since the actual number of underlying topics is unknown and there is no definite and efficient approach to accurately estimate it, the inferred topics of any topic model do not always represent meaningful themes. This dissertation presents the first automated unsupervised analysis of LDA models to rank the topics based on their semantic significance and, eventually, identify and distinguish junk topics from legitimate ones. The basic idea consists of measuring the distance between topic distribution and some "junk distribution." In particular, three definitions of "junk distribution" are introduced, and a variety of metrics are used to compute the distances, from which an expressive figure of the semantic significance of the topics is implemented using a 4-phase Weighted Linear Combination approach.

Non-Degree Open House: June

Wednesday, June 17, 2009,
6:00PM-8:00PM,
Fairfax Campus, Engineering Building, 2nd FL Lobby area

The Volgenau School of IT & Engineering will offer a series of Non-Degree Open Houses this summer where prospective students interested in taking graduate coursework for the Fall 2009 term an incredible opportunity to:

* Learn about our graduate programs * Apply as a non-degree student * Obtain an 'on-the-spot' admissions decision * Register and pay for classes This is a one-stop opportunity to get started on the road towards your graduate education!

Eligibility Disclaimer: Individuals seeking or holding F1 or J1 visas are not eligible for non-degree status, but may apply for any of our degree programs.

Be sure to bring the application materials indicated on our Non-Degree Open House requirements checklist.

RSVPs are required: http://volgenau.ite.gmu.edu/graduateresearch/responseform/

Find Directions/Map: http://coyote.gmu.edu/map

Parking Information/Cost: http://parking.gmu.edu/visitors.htm

If you have questions about this event please contact us at itegrad@gmu.edu.

Non-Degree Open House: July

Wednesday, July 15, 2009,
6:00PM-8:00PM,
Fairfax Campus, Engineering Building, 2nd FL Lobby area

The Volgenau School of IT & Engineering will offer a series of Non-Degree Open Houses this summer where prospective students interested in taking graduate coursework for the Fall 2009 term an incredible opportunity to:

* Learn about our graduate programs * Apply as a non-degree student * Obtain an 'on-the-spot' admissions decision * Register and pay for classes

This is a one-stop opportunity to get started on the road towards your graduate education!

Eligibility Disclaimer: Individuals seeking or holding F1 or J1 visas are not eligible for non-degree status, but may apply for any of our degree programs. Be sure to bring the application materials indicated on our Non-Degree Open House requirements checklist.

RSVPs are required: http://volgenau.ite.gmu.edu/graduateresearch/responseform/

Find Directions/Map: http://coyote.gmu.edu/map

Parking Information/Cost: http://parking.gmu.edu/visitors.htm

If you have questions about this event please contact us at itegrad@gmu.edu.

GRAND seminar: Secure Computation on Encrypted Database

Thursday, July 23, 2009,
11:00AM,
ENGR 4201
David Wai-lok Cheung

Head of Department of Computer Science
Director of the Center for E-commerce Infrastructure Development (CECID)
University of Hong Kong

Abstract

Google and Amazon have turned their huge infrastructure into a cloud computing environment and are aggressively recruiting businesses to run applications on their platforms. Many top-tier IT vendors have been promoting cloud computing as a new service model. However, serious concern has been raised about the security and privacy on such a service platform. Corporate users would need to protect their data running on the cloud computing platform from the “untrusted” service provider. Unfortunately, traditional encryption methods that aim at providing “unbreakable'' protection are often not adequate because they are not designed to support applications to be executed on the encrypted data. In this seminar, the general problem of computing on encrypted data will be discussed. The key issue is how to balance the need on security and the requirement to perform computation. As a case study, the problem of k-nearest neighbor (kNN) computation will be used to illustrate a new model of secure computation on encrypted database.

Short Bio

Professor David Wai-lok Cheung is the Head of Department of Computer Science and Director of the Center for E-commerce Infrastructure Development (CECID) in The University of Hong Kong. He conducts research in database, data mining and e-commerce technologies. He was the recipient of the HKU Outstanding Researcher Award. Most recently, he received the Distinguished Contribution Award in the 2009 Pacific-Asia Data Mining and Knowledge Discovery Conference. He was the program chairman of the 2001 and 2005 Pacific-Asia Knowledge Discovery and Data Mining Conferences, the conference chairman of the 2007 PAKDD Conference. Most recently he is the conference co-chair of the 2009 CIKM Conference. Concerning applied research, he has received more than HK$60M grants from the Hong Kong Innovation and Technology Commission. His team at HKU has developed open-source ebXML gateway, which has been used by developers in e-commerce and logistics from more than 80 countries. The open-source product has received awards in various prominent competitions, including the Hong Kong 2004 IT Excellence Awards, the 2004 Asia-Pacific ICT Awards, and the 2005 Linux Business Awards.

Non-Degree Open House: August

Wednesday, August 12, 2009,
6:00PM-8:00PM,
Fairfax Campus, Engineering Building, 2nd FL Lobby area

The Volgenau School of IT & Engineering will offer a series of Non-Degree Open Houses this summer where prospective students interested in taking graduate coursework for the Fall 2009 term an incredible opportunity to:

* Learn about our graduate programs * Apply as a non-degree student * Obtain an 'on-the-spot' admissions decision * Register and pay for classes

This is a one-stop opportunity to get started on the road towards your graduate education!

Eligibility Disclaimer: Individuals seeking or holding F1 or J1 visas are not eligible for non-degree status, but may apply for any of our degree programs. Be sure to bring the application materials indicated on our Non-Degree Open House requirements checklist.

RSVPs are required: http://volgenau.ite.gmu.edu/graduateresearch/responseform/

Find Directions/Map: http://coyote.gmu.edu/map

Parking Information/Cost: http://parking.gmu.edu/visitors.htm

If you have questions about this event please contact us at itegrad@gmu.edu.

PhD Dissertation Defense: Efficient Resource Management for Heterogeneous Devices Accessing Internet Streaming Content

Wednesday, August 19, 2009,
10:00AM - 12:00PM,
Engineering Building, Suite 4201
DONGYU LIU

CS PhD Candidate

Abstract

The volume of streaming media content has surged on the Internet in recent years. In parallel, with the technology advancement in wireless and third generation (3G) networks, more and more Internet users today use their mobile devices, such as smart phones and PDAs, to access Internet streaming content. However, compared with desktops, mobile devices normally have different display sizes, color depths, bandwidth capacities, CPU speeds, and battery capacities, which are termed Multiple Dimensional Heterogeneity (MDH). Due to the MDH problem, these mobile devices normally cannot directly display the data streamed to desktops. A few solutions based on Multiple Description Coding (MDC) or Scalable Coding (SC) have been proposed, but none of these solutions is practical due to their various limitations.

In this dissertation, novel efficient resource management and transcoding schemes are investigated to address the MDH problem in both Internet on-demand and live streaming systems. First, in on-demand streaming systems, meta-data caching is leveraged during the online transcoding to reduce the overall CPU consumption. A model is further constructed to study the tradeoff between the CPU and the storage consumption for different meta-caching schemes under various conditions. Accordingly, an adaptive scheme that always outperforms existing schemes is proposed. Second, in live streaming systems, the idle peer computing cycles are leveraged for efficient online transcoding based on an additional transcoding overlay constructed by participating transcoding peers. This proposed scheme effectively explores the tradeoff between the CPU and the bandwidth resources. Third, considering resource contributions made by heterogeneous devices and their limited available resources, e.g., battery capacity, we propose a dynamic rotation scheme in both streaming and meta-transcoding overlays to balance the resource consumption among different peers, aiming to improve the fairness among peers and the overall system robustness. Fourth, a general theoretical framework is constructed to maximize the resource utilization in heterogeneous streaming systems. Accordingly, a distributed algorithm is proposed to achieve near-optimal performance with acceptable control overhead.

Orientation: New Graduate Student

Wednesday, August 26, 2009,
5:30PM,
Research I Room 163

Abstract

The Volgenau School of Information Technology and Engineering is hosting its annual special orientation event to welcome all newly admitted graduate students.

Although orientation is not mandatory, it is strongly recommended that all degree-seeking graduate students plan to attend. Essential information regarding university services for graduate students and program information from the Computer Science departments will be provided. Most importantly, this is your opportunity to meet your peers, the administrative and academic staff members who will assist you during the pursuit of your graduate course work and degree.

RSVP Please RSVP online for orientation by visiting: http://ite.gmu.edu/graduateresearch/responseform/

Please contact the Office of Graduate Admissions & Enrollment Services at itegadm@gmu.edu should you have any questions.

Orientation: GTA

Thursday, August 27, 2009,
12:00-1:30PM,
ENGR 4201

Orientation: PhD Students

Thursday, August 27, 2009,
11:00-12:30PM,
ENGR 4201

Computer Science: Mandatory GTA Workshop

Thursday, September 03, 2009,
10:30am - 12:30pm,
ENGR 4201

SANG Seminar: VirusMeter: Protecting Your Cellphone from Spies

Friday, September 04, 2009,
12:00-1:30pm,
ENGR 4201
Lei Liu

URL: http://cs.gmu.edu/~sqchen/SANG/

Abstract

Due to the rapid advancement of mobile communication technology, mobile devices nowadays can support a variety of data services that are not traditionally available. With the growing popularity of mobile devices in the last few years, attacks targeting them are also surging. Existing mobile malware detection techniques, which are often borrowed from solutions to Internet malware detection, do not perform as effectively due to the limited computing resources on mobile devices.

In this work, we propose VirusMeter, a novel and general malware detection method, to detect anomalous behaviors on mobile devices. The rationale underlying VirusMeter is the fact that mobile devices are usually battery powered and any malicious activity would inevitably consume some battery power. By monitoring power consumption on a mobile device, VirusMeter catches misbehaviors that lead to abnormal power consumption. For this purpose, VirusMeter relies on a concise user-centric power model that characterizes power consumption of common user behaviors. In a real-time mode, VirusMeter can perform fast malware detection with trivial runtime overhead.

Speakers' Bio

Lei Liu is a Ph.D. student in Computer Science Department of George Mason Univesity. His research interests incude system security, network application, operating systems and software engineering. Before pursuing his Ph.D. degree, he has several years' experience in IT industry.

SANG seminar: Business-Oriented Autonomic Load Balancing for Multitiered Web Sites

Thursday, September 10, 2009,
03:00-04:30PM,
Engineering Building, Room 4201
John Ewing

PhD student in the Department of Computer Science

Abstract

This presentation will discuss the application of autonomic computing to load-balancing for a multi-tiered auction web site. Autonomic computing systems are able to adapt to changing environments (such as changes in the workload intensity or component failures) in a way that preserves high-level operational goals, such as service level objectives. The autonomic load balancer of the web site divides the bottleneck server tier into clusters, each of which is dedicated to a certain priority class of users. The autonomic load balancer dynamically adjusts resource allocations to the clusters in an effort to maximize a utility function based on response time and bid throughput. To reduce switching costs, the load balancer also considers load balancing policies that redirect a percentage of requests intended for one cluster to a different cluster. To make allocation and policy decisions, the autonomic load balancer uses an efficient heuristic search to explore a utility landsc ape generated by the predictions of an analytic performance model. Another key contribution presented here is a novel method for the random generation of realistic stress tests. Using this stress test method, the autonomic load balancer is assessed against both a round-robin load balancing approach and a dedicated cluster approach.

Speaker's bio

John Ewing is currently a PhD student in the Computer Science Department at George Mason University. From 2001 to 2005, John worked at the Defense Information Systems Agency conducting capacity planning studies and performance troubleshooting of large, distributed software systems with millions of users. From 1999 to 2001, John worked as a government IT contractor analyzing performance of computer systems and developing prototype software. John received his Masters in computer science from the Illinois Institute of Technology in 2003 and his Bachelors in chemistry from the University of Richmond in 1997.

GRAND Seminar: Digital Geometry and 3D Imagery: Topological methods

Friday, September 11, 2009,
12:00 Noon,
ENGR 4201
Chen Li

Associate Professor
Department of Computer Science and Information Technology
University of the District of Columbia

Abstract

In this talk, we will focus on digital geometry and topology methods of 3D image processing and computer vision. We first introduce connectivities in digital space and algorithms for component extraction. Then we discuss digital surfaces and manifolds including their recognition procedures. We will also present fast algorithms to compute invariants such as genus and homology groups for 3D objects in 3D space. Finally, we will discuss the methods for high dimensional cloud point data processing including persistent analysis and its relationship to manifold learning.

Short Bio

Dr. Li Chen is an associate professor of computer science at the University of the District of Columbia. He is currently working on problems in image segmentation algorithms, complexity analysis of algebraic groups, and the relationship between finite elements and gradually varied fitting. Li Chen received his BS, MS, and Ph.D. all in CS from Wuhan University (1982), Utah State University (1995), and the University of Bedfordshire (Luton, UK, 2001), respectively. His work includes: 1) The definitions of digital surfaces and manifolds; 2) Digital surface points classifications; and 3) The lambda-connected search algorithm for image segmentation, a dual-technique to threshold segmentation, the most popular segmentation method.

SANG seminar: Quantification of Computer Security: Some Case Studies

Thursday, September 17, 2009,
03:00-04:30PM,
Engineering Building, Room 4201
Prof. Michel Cukier

an Associate Professor of Reliability Engineering at
the University of Maryland, College Park.

Abstract

In this talk, I will discuss several case studies related to the quantification of computer security. In particular, I will present results obtained from a test-bed deployed at the University of Maryland to collect attack data using target computers (i.e., honeypots). The data provided the evidence needed in the following four research threads. (1) Within the security community, scans are usually considered as precursors to an attack. However, few studies quantified the validity of this hypothesis. We introduced packet-counting models to identify TCP scans and attacks and applied these models to analyze malicious traffic towards honeypots. (2) We developed a methodology for determining the characteristics that separated attacks most efficiently. A comparison between the analysis of attack messages and the outcome of a clustering algorithm indicated the efficiency of the characteristic. (3) We built a profile of attacker behavior following a remote compromise by looking for specific actions taken by attackers. The results represent solid statistical evidence to support widely held beliefs about post-compromise attacker behavior. (4) We compared malicious traffic originating inside UMD with that originating outside UMD. We showed that internal malicious traffic often contained different malicious content compared to that of external traffic.

Speaker's bio

Prof. Michel received a degree in physics engineering from the Free University of Brussels, Belgium, in 1991, and a doctorate in computer science from the National Polytechnic Institute of Toulouse, France, in 1996. From 1996 to 2001, he was a researcher at the University of Illinois, Urbana-Champaign.

He joined the University of Maryland in 2001 as Assistant Professor. His research covers dependability and security issues. His latest research focuses on the empirical quantification of computer security. He has published over 60 papers in journals and refereed conference proceedings in those areas.

GRAND Seminar: From recognizing biological sequences, to identifying search keywords: A feature generation framework

Tuesday, September 22, 2009,
12:00pm,
Engineering Building, Room 4201
Rezarta Islamaj

Research Fellow
National Center for Biotechnology Information (NCBI)
NIH

Abstract

The set of attributes or features selected to model an entity is very important for correct classification. In this talk I will present an integrated process, which I refer to as feature generation. This method allows the user to construct informative features based on domain knowledge, and to search a large space of potential features effectively. I applied this approach to the problem of splice-site prediction and obtained new predictive models for these biological signals for two different organisms. These models have achieved significant improvements in accuracy over existing, state-of-the-art approaches. In each case, the identified sets of features were used to discover biologically interesting motifs. They are available to the public through an easy-to-use website, SplicePort (http://www.spliceport.org). Spliceport can be used to predict new splice sites from user-input sequences, and to browse the whole collection of features for biologically significant signals. I also applied this approach to the problem of keyword identification for effective document retrieval. The automatic identification of ?clickable? words in the title and abstract of articles is of central importance in improving the retrieval quality of the search engine. It is also important to authors as it increases the chances that their article will get better visibility. PubMed (http://www.ncbi.nlm.nih.gov/PubMed), a free Web service provided by the U.S. National Library of Medicine, provides daily access to over 19 million biomedical citations for millions of users. The current retrieval algorithm in PubMed finds all the articles that match the terms in the user query and presents them in reverse chronological order. I studied PubMed log data for the clickthrough activities of users after they have issued a query. Linking the query terms to the clicked articles, I built a novel machine learning model that identifies "keywords" that are preferred by users to access a particular article.

Short Biography

Dr. Rezarta Islamaj received her Ph.D. degree in Computer Science from University of Maryland at College Park in 2007. Her research focused on applying machine learning and data mining approaches to computational biology problems. Specifically she worked on construction, selection and discovery of appropriate motifs to model biological signals for accurate classification and prediction.

CS Seminar: Faculty Research Overview

Wed, Sept 23, 2009,
2:00 PM,
Engineering Building, Room 4201
Prof. Duric, Prof. Rangwala, Prof. Kosecka

Abstract

We will have the first monthly seminar reviewing research activities and projects of computer science faculty. Professors Duric, Rangwala and Kosecka will each speak for 15 minutes about their current interests and projects, followed by questions and discussions.

Prof. Zoran Duric: Understanding Human Movement (Area: Computer Vision, Robotics, AI)

Prof. Huzefa Rangwala: Learning for Metagenomics and Structural Bioinformatics (Area: Bioinformatics, Data Mining)

Prof. Jana Kosecka: Virtual Travel using Image Based Models (Area: Computer Vision, Robotics, AI)

SANG seminar: On the Effectiveness of Low Latency Anonymous Network in the

Friday, September 25, 2009,
12:00 - 01:30pm,
Engineering Building, Room 4801
Jing Jin

PhD Student in the Department of Computer Science

Abstract

An anonymous network provides services to hide the correspondence between its incoming and outgoing messages. Timing attack is a significant challenge for anonymous networks that support interactive and low-latency applications. We introduce a novel metric that can quantitatively measure the practical effectiveness (i.e. anonymity) of all anonymous networks in the presence of timing attack. Our metric is based on a novel measurement of the distortion of the packet timing between the incoming and the outgoing flows to and from the anonymous network and it uses wavelet based analysis to measure the variability of the distortion. To the best of our knowledge, our approach is the first practical method that can quantitatively measure the packet timing distortion between flows that may have gone through such transformations as flow mixing/spliting/merging, adding chaff, packet dropping.

To validate our anonymity metric, we have conducted real-time timing attacks on various deployed anonymous networks such as Tor, anonymizer.com and have used the timing attack results as the ground truth for validating our anonymity metric. We have found strong correlation between our anonymity metric and the timing attack results. Our metric measurements and timing attack results show that the circuit rotation in Tor network could significantly increase its resistance to timing attack at the cost of more timing disturbances to the normal users. In addition, we have found that adding constant rate chaff (i.e. cover traffic) has diminishing effect in anonymizing packet flows.

Bio

Jing Jin is a Ph.D. student in Computer Science Department of George Mason University. Her research interests include system and networking security, software engineering.

GRAND Semianr: Reconstruction, Localization and Semantic Parsing of Urban Scenes

Tuesday, September 29, 2009,
12:00 Noon,
ENGR 4801
Jana Kosecka

Associate Professor
Department of Computer Science
George Mason University

Abstract

I will describe a method for 3D reconstruction of urban scenes, which along with the ego-motion estimation and depth map fusion constitutes a robust, purely vision based method for capturing 3D models of urban environments. I will also talk briefly about some on-going work on semantic segmentation of urban areas using spatial co-occurrence of visual words and 3D geometry. I will show some preliminary results on challenging urban scenes with varying viewpoints and large number of categories appearing simultaneously.

Software Engineering Seminar: A Modeling Language for Activity-Oriented Composition of Service-Oriented Software Systems

Wednesday, September 30, 2009,
12:00 - 01:00pm,
Engineering Building, Room 4801
Naeem Esfahani

PhD Student
Department of Computer Science

Abstract

The proliferation of smart spaces and emergence of new standards, such as Web Services, have paved the way for a new breed of software systems. The functional and QoS requirements of such software systems are often not known a priori at design-time, and even if they are, they may change at run-time. Unfortunately, the majority of existing software engineering techniques rely heavily on human reasoning and manual intervention, making them inapplicable for automatic composition of such software systems at run-time. Moreover, these approaches are primarily intended to be used by technically knowledgeable software engineers, as opposed to domain users. In this talk, we present Service Activity Schemas (SAS), an activity-oriented language for modeling software system’s functional and QoS requirements. SAS targets service-oriented software systems, and relies on an ontology to provide domain experts with modeling constructs that are intuitively understood. SAS forms the centerpiece of a framework intended for user-driven composition and adaptation of service-oriented software systems in a pervasive setting (SASSY).

Bio

Naeem Esfahani is a PhD candidate in Computer Science Department, Volgenau School of Information Technology and Engineering. He got his bachelor’s degrees on Computer Engineering with major of Software Engineering from University of Tehran, Tehran, Iran. He also holds a Master of Science degree in Computer Engineering with major of Software Engineering from Sharif University of Technology, Tehran, Iran. His current research mainly focuses on Software Architecture, Autonomic Computing, Model Driven Development, Pervasive Systems, and Software Development Processes.

Engineering Building: Grand Opening

Friday, October 02, 2009,
3-5 PM,
Engineering Building Courtyard & Atrium

Abstract

Come visit the Robotics Lab (Room 2201) and the Visual Computing Lab (Room 4405) for demos and see posters displaying research by CS faculty in Room 2901.

SANG seminar: Exploitation and Threat Analysis of Open Mobile Devices

Friday, October 16, 2009,
12:00 - 01:30pm,
Engineering Building, Room 4201
Lei Liu

PhD Student in the Department of Computer Science

Abstract

The increasingly open environment of mobile computing sys- tems such as PDAs and smartphones brings rich applica- tions and services to mobile users. Accompanied with this trend is the growing malicious activities against these mobile systems, such as information leakage, service stealing, and power exhaustion. Besides the threats posed against individual mobile users, these unveiled mobile devices also open the door for more serious damage such as disabling criti- cal public cyber physical systems that are connected to the mobile/wireless infrastructure. The impact of such attacks, however, has not been fully recognized.

In this talk, we show that mobile devices, even with the state-of-the-art security mechanisms, are still vulnerable to a set of carefully crafted attacks. Taking Linux-based cellphones as an example, we show that this vulnerability not only makes it possible to attack individual mobile devices such as accessing unauthorized resources, disabling predefined security mechanisms, and diverting phone calls, but also can be exploited to launch distributed denial-of-service attacks against critical public services such as 911.

Bio

Lei Liu is a Ph.D. student in Computer Science Department of George Mason Univesity. His research interests incude system security, network application, operation systems and sogrware engineering. Before pursuing his Ph.D. degree, he has several years' experience in IT industry.

Software Engineering Seminar: Tulips, Potatoes, Apples, ISO 9001 and the CMMI

Friday, October 16, 2009,
02:00-03:00pm,
Engineering Building, Room 4705
Nelson Perez

President of Sierra's Edge Inc.

Abstract

There are many influences that come into play when creating new processes or improving existing processes. One can be a result of processes or experiences people bring into an organization from the outside. Others include the influence of existing company culture; or having to be in compliance with a particular best practice model like the CMMI or standard like ISO 9001; or even what one comes across in literature, on the internet or even at forums such as conferences and SPIN meetings.

We are continually being influenced to change and improve the way we do things. This presentation shows an example of how influences come in play to create and improve an example process. In this case, a verification and validation (V&V) process for a software application that has to meet nuclear quality standards (NQA-1 and IEEE). We will explore how this example process comes into existence under various influences, including how it is affected by personal experience, different best practice models, cultural resistance, and continuous improvement.

In the end, the most successful processes are those that incorporate multiple influences to create a best fit solution. Not unlike how plants adapt and evolve to successfully survive in a constantly changing world. Plants do this naturally by pollination, but sometimes plants receive outside “help” from other sources. These pollination vectors work together to influence the resulting offspring across multiple generations. In the end, aren’t we all doing the same thing, pollinating and spreading our ideas and controlling the evolution of processes? This presentation will show how us human pollinators work together to bring about the continuing adaptations needed for successful process improvement.

Bio

Nelson Perez is president of Sierra’s Edge, Inc. a process development and improvement consultant. With over 25 years of work experience, he has worked the entire product development life cycle and held numerous management and engineering positions on such high visibility programs as the B2 Stealth Bomber; NASA Space Shuttle; and National Missile Defense. He has worked a variety of other domains including security systems, information systems, computer integrated manufacturing, wireless communication, commercial software development, aircraft systems, and non-destructive testing. He has co-authored 1 NASA-related patent, has authored several papers and is an active presenter at conferences, software process improvements network (SPIN) meetings and other venues. Perez consults with companies in management, technology innovation and process and business improvement including ISO 9001:2000, CMMI, and ITIL v3.

GRAND Seminar: Toward Physical Universal Constructors: Materials, Processes, Modules, and System

Tuesday, October 20, 2009,
12:00 Noon,
ENGR 4201
Matt Moses

Ph.D. Student
Department of Mechanical Engineering
Johns Hopkins University

Abstract

More than 50 years ago the mathematician John von Neumann introduced the idea of a Universal Constructor - a machine that could build anything described to it, including copies of itself. Initially von Neumann investigated a physically realistic design, often called the "Kinematic Model", before abandoning it to focus on a more abstract formulation of the problem. Together with mathematician Stanislaw Ulam, von Neumann developed the concept of Cellular Automata (CA) as a mathematical tool for rigorous study of Universal Constructors. The CA model has seen a fair amount of success, and is currently an active area of study. Progress in Kinematic Universal Constructors has lagged behind, although the field has seen a resurgence in activity in recent years. This talk begins with a brief historical overview of Universal Constructors and Self-Replicating Machines. We then present some recent results from our lab: a network of materials and fabrication processes designed to facilitate self-replication, a set of universal electromechanical modules, and some larger scale system designs. The talk is concluded with a discussion of some open questions that may be of interest to those with a Computer Science background.

Short Bio

Matt Moses is currently a doctoral student in Professor Greg Chirikjian's lab at the Department of Mechanical Engineering, Johns Hopkins University. Mr. Moses holds a M.S degree in Mechanical Engineering from University of New Mexico, and from 2001-2005 worked as an engineer for General Dynamics Robotic Systems in Westminster, Maryland. Mr. Moses's research interests include self-replicating and universal-constructing machines, biologically-inspired robotics, and dynamic running and locomotion in robot vehicles.

CS Seminar: Faculty Research Overview

Wednesday, October 21, 2009,
2:00 - 3:00 PM,
Engineering Building, Room 4201
Professors Brodsky, Wechsler, Sood

Abstract

We will have the second monthly seminar reviewing research activities and projects of computer science faculty. Professors Brodsky, Wechsler and Sood will each speak for 15 minutes about their current interests and projects, followed by questions and discussions.

Prof. Brodsky: Decision-Guidance Systems

How would you design a computer systems to interract and guide human decision makers to move a complex process toward most desirable outcomes? How would you help find the best course of action in emergency, best business transactions, or recommend public policies guided by most positive outcomes? This talk will tell you about recent advances in decision-guidance systems, and exciting research opportunities in this area.

Prof. Wechsler: Robust Biometrics

Robust biometrics is about reliable authentication and identity management despite less than (1) perfect sensors, (2) missing data, and (3) corrupt information.

Prof. Sood: Beyond Prevention and Detection – Intrusion Tolerance

Our response to cyber security challenges is a time based intrusion tolerance system. We call our approach Self Cleansing Intrusion Tolerance (SCIT). SCIT servers are focused on limiting the losses that can occur because of an intrusion. To achieve this goal we limit the exposure time of the server to the internet. In the SCIT approach our goal is sub-minute exposure time for servers without service interruption.

Joint CS/Volgenau School Seminar: Improving and Securing Mobile Internet Accesses

Wednesday, October 21, 2009,
11:30 AM,
Johnson Center, Gold Room
Prof. Songqing Chen

Abstract

With the rapid advancement of mobile communication technology, increasingly more users rely on mobile devices, such as smart phones and PDAs, to connect to the Internet, However, this technology trend and wide user accessibility have presented new challenges to the current practice from various aspects, ranging from improved Quality of Service to better security and privacy demanded by mobile users.

In this talk, I will discuss some of these challenges and my research solutions to address them, in addition to an overview of my research. Specifically, I will discuss how to efficiently address the multiple dimensional heterogeneity problem in both on-demand and live streaming services. In order to solve increasingly serious problems of mobile malware on mobile devices, I will also present a novel technique to detect mobile malware in order to protect mobile users.

Bio:

Dr. Songqing Chen is an Assistant Professor of Computer Science in the Volgenau School of IT & Engineering. He received his Ph.D. from the College of William and Mary, 2004. His research interests mainly focus on design, analysis, and implementation of algorithms and experimental systems in distributed and networking environments, particularly, in the areas of Internet content delivery systems, Internet measurement and modeling, operating systems and system security, distributed systems, and high performance computing. Dr. Chen has served on various system and networking conference TPCs, including IEEE INFOCOM, ICDCS, ACM MM, and WWW. Currently, he is serving as the vice co-chair of ICDCS 2010 for the Internet/network protocol track. His research has been supported by NSF, HP Labs, and AFOSR. He is a recipient of the NSF CAREER Award and AFOSR YIP Award.

Software Engineering Seminar: Mutation Testing: Towards industrial application

Monday, October 26, 2009,
12:00 - 01:00pm,
Engineering Building, Room 4201
Pedro Reales Mateo

Abstract

Since the 1970s, mutation has been a testing technique widely used by researchers, who have applied it mainly for the validation of test suites, as well as for the validation of strategies for test case or test data generation. In our opinion, mutation is today mature enough for its application in industrial environments. Although the three main steps of mutation (mutant generation, test case execution and result analysis) may be quite costly, research has produced a good set of results that make possible the application of all this knowledge in industry. This work reviews some of this cost problems and the solutions proposed by the research and present a tool, Bacterio, which try to combine this solutions to facilitate the application of the mutation in the industry.

Bio

Pedro Reales has a MSc degree in Computers Science by the university of Castilla-La Mancha and has a Master in Avanced Computers Technologies by the university of Castilla-La Mancha. Currently, he is an active member of the Alarcos Research Group and is an Interchange Visitor in the George Mason University. His research areas are related to the automation software testing tasks.

GRAND seminar: Role of promiscuous binding and intrinsic disorder

Tuesday, October 27, 2009,
12:00pm,
Engineering Building, Room 4201
Anna Panchenko

Associate Investigator
NCBI, NIH

Abstract

Cellular processes are highly interconnected and many proteins are shared in different pathways. Some of these shared proteins or protein families may interact with diverse partners using the same interface regions. Analysis of such regions is essential for understanding the mechanisms of specific molecular recognition of multiple diverse partners. We find that only 5 percent of protein families in the structure database have multibinding interfaces, and they do not show any higher sequence conservation compared with the background interface sites. We highlight several important functional mechanisms utilized by multibinding families. Promiscuous interactions can also be studied by using our recently developed IBIS server (Inferred Biomolecular Interaction Server, http://www.ncbi.nlm.nih.gov/Structure/ibis/ibis.html) which analyzes and annotates the interaction partners and locations of binding sites in proteins. It has been suggested that intrinsic disorder contributes to the ability of some proteins to interact with multiple partners as folding of disordered proteins into ordered structures may occur upon binding to their specific partners.We performed a large-scale study of intrinsically disordered regions in proteins and protein complexes. In accordance with the conventional view that folding and binding are coupled, in many of our cases the disorder-to-order transition occurs upon complex formation and can be localized to binding interfaces. Moreover, analysis of disorder in protein complexes depicts a significant fraction of intrinsically disordered regions, with up to one third of all residues being disordered. We find that the disorder in homodimers, especially in symmetrical homodimers, is significantly higher than in heterodimers and offer an explanation for this interesting phenomenon. The fascinating diversity of roles of disordered regions in various biological processes and protein oligomeric for ms shown in our study is an important subject for future endeavors in this area.

Host

Amarda Shehu

Joint CS/Volgenau School Seminar: Exploring the Maze of MIX Networks and Malwares

Wednesday, October 28, 2009,
11:30 AM,
Jajodia Auditorium, Engineering Bldg
Prof. Xinyuan (Frank) Wang

Assistant Professor, Dept. of Computer Science, GMU

Abstract

The concept of MIX is fundamental to all anonymous communication networks, and almost all existing anonymous networks used traffic mixing and transformation to achieve anonymity. It has long been believed that flow mixing and transformations would effectively disguise network flows, thus achieve good anonymity. In the first half of this talk, I will overview my work in investigating the fundamental limitations of flow mixing and transformation in achieving anonymity. I will describe how active flow watermarking in packet timing could transparently make a sufficiently long flow uniquely identifiable. I will also discuss how active flow watermarking can be applied to track anonymous, peer-to-peer VoIP calls on the Internet. In the second half of my talk, I will overview my work in malware analysis and defense.

Bio:

Xinyuan Wang is currently an Assistant Professor in the Department of Computer Science at George Mason University. He received his BS and MS in Computer Science from Peking University and Chinese Academy of Space Technology respectively. He received his PhD in Computer Science from North Carolina State University in 2004 after years of professional experience in the networking industry. His main research interests are around computer network and system security including malware analysis and defense, attack attribution, anonymity and privacy, VoIP security, digital forensics. He has developed the first inter-packet timing based packet flow watermarking scheme that is provably robust against timing perturbation. He has first demonstrated that it is feasible to track encrypted, anonymous peer-to-peer VoIP calls on the Internet. In his later work, he has demonstrated the fundamental limitations of existing low-latency anonymous communication systems in the presence of timing attack, and developed the first practical attack that has penetrated the Total Net Shield the ultimate solution in online identity protection of www.anonymizer.com with less than 11 minutes worth of Internet traffic. Xinyuan Wang received the 2009 NSF Faculty Early Career Development (CAREER) Award, the 2009 GMU CS Outstanding Faculty Research award, and the 2004 ACM Graduate Student Research Competition Grand Finals Award (Third Place).

GRAND seminar: Computing structural changes in proteins

Tuesday, November 03, 2009,
12:00pm,
Engineering Building, Room 4201
Nurit Haspel

Assistant Professor
Department of Computer Science
University of Massachusetts Boston

Abstract

Proteins are involved in virtually every process and aspect in life ? from the flexing of our muscles to our immune system response. It is widely accepted that proteins are dynamic molecules with generally well-defined three-dimensional structures, and that understanding the structure and dynamics of proteins is crucial for understanding their function and the processes they mediate. Various computational methods exist for modeling and simulating protein structure and dynamics, but several traditional methods are limited due to the large amount of calculations involved. This talk will present a number of methods for searching the conformational space of proteins at various time and space scales. On the one end of the spectrum, Molecular Dynamics calculations are used for detailed analysis of local structural changes in proteins. On the other end, robotics-inspired search techniques are used to characterize the structure and dynamics of proteins by representing them using a mechanistic/geometric models subject to physics constraints.

Bio

Nurit Haspel is an assistant professor in the department of Computer Science at the University of Massachusetts in Boston. She received her BSc. in Chemistry and Computer Science from Tel Aviv University and her Masters and Ph.D in Structural Bioinformatics (Computer Science) from Tel Aviv University in 2007. She did her postdoctoral research in the department of Computer Science at Rice University in Houston, TX. Her research involves developing and applying computational methods to explore the structure and dynamics of protein molecules. Some applications include the computational design of nano-structures, applying computational simulations and search techniques to understand aspects of the human immune system inhibition and developing algorithms taken from Robotics and graph theory to simulate large scale structural changes in protein complexes.

Joint CS/Volgenau School Seminar: A Day in the Life of an Access Controller on the WWW

November 4, 2009,
11:30 AM,
Jajodia Auditorium, Engineering Bldg
Duminda Wijesekera

Associate Professor, Computer Science Department

Abstract

We still wish to govern the accesses to resources residing on large, loosely coupled and mostly uncoordinated distributed systems such as the WWW. In order to do so, we wish to create access control frameworks that are sufficiently generic so that they can be used by multiple application domains. In order to do so, one needs to be able to define all relevant entities in a manner that is name-space independent, but yet customizable as needed. In addition, the framework needs to be able to interpret security policies and provide decisions that have to be computed amidst (distributed) divergences and failures. Additionally, enforcing some decisions made by such policy frameworks may change the policies themselves, thereby requiring the transactional and locking mechanisms in order enforce the rendered decisions. Yet, these policy frameworks, being constructed out of a collection of communicating processes, have to evolve from their birth to death in a manner that reflects their operational environment in a policy consistent manner. This talk describes some work that has been conducted in cooperation with a group of individuals on enhancing the eXtensibe Access Control Meta Language (XACML) and its enforcement framework.

Software Engineering Seminar: Architectural Patterns for Decentralized Self-Adaptive Systems

Monday, November 09, 2009,
12:00 - 01:00pm,
Engineering Building, Room 4201
Danny Weyns

Abstract

Self-adaptability has been proposed as an effective approach to tackle the increasing complexity of constructing and managing contemporary software systems. Self-adaptability endows a system with the capability to adapt itself to changes in its environment and user requirements. Several researchers have argued that software architecture provides the right level of abstraction and generality to deal with the challenges of self-adaptability. One of the major challenges in self-adaptive systems is dealing with distribution and decentralization. Decentralized control is crucial for quality requirements such as openness and scalability. Over the past 8 years, we have been studying decentralized architectures for realizing self-adaptability based on multi-agent systems (MAS). A MAS architecture structures the software in a number of interacting autonomous entities (agents) that cooperatively realize the system goals. Agents flexibly adapt their behavior and interactions to dynamics in the system or its environment. In the course of designing and building various MAS applications, we derived several architectural patterns that provide generic solution schemes for recurring design problems. In this talk, I will zoom in on a number of these patterns and illustrate them with practical examples.

Bio

Danny Weyns is a post-doctoral researcher at the Katholieke Universiteit Leuven, Belgium. His main research interests are in software architecture, self-adaptive systems, multi-agent systems, and middleware for decentralized systems. Danny is currently visiting researcher at Valoria Lab of the Université de Bretagne-Sud, France where he works with Prof. Flavio Oquendo on formal modeling of dynamic software architectures of decentralized systems.

Software Engineering Seminar: Continuous Learning for Self-Adaptive Software Systems

Monday, November 09, 2009,
12:00 - 01:00pm,
Engineering Building, Room 4201
Jesper Andersson

Abstract

Recent computing trends, such as pervasiveness and mobility of systems, increase the software complexity to a point where the software itself must adapt at run-time. In order to meet functional and quality requirements in a setting of ever changing conditions and constraints, self-adaptation is one possible answer. Adaptive system design, development, and maintenance add additional challenges for software engineers. In addition to the complexity of static system engineering, systems must know when to adapt, where to adapt, and how to adapt. Acquiring knowledge to answer these questions is a challenging task and the problem becomes even more complex as the systems, and number of conditions and constraints grow. In this talk we present and exemplify work in progress on a dual learning process with architecture support that combines off-line and on-line mechanisms. Off-line, the system is simulated or subject for controlled executions. The collected information is used to derive configuration knowledge, i.e., configuration rules. We realize that off-line learning will never be 100 percent correct or complete in the general case. We address this issue by employing on-line learning strategies for gradually tuning/extending/replacing configuration knowledge.

Bio

Jesper Andersson is an assistant professor at Växjö University in Sweden. His main research interests are in software architectures, end-user programming and reuse for self adaptive systems development and self adaptation in distributed architectures.

GRAND seminar: United we stand, divided we fall: Integrating Continuous Robot Motion

Tuesday, November 10, 2009,
12:00 - 01:00pm,
Engineering Building, Room 4201
Erion Plaku

Postdoctoral Fellow
Laboratory for Computational Sensing and Robotics
Johns Hopkins University

Abstract

Research in robotics has focused since its inception towards increasing the ability of robots to plan and act on their own in order to complete assigned high-level tasks.

Toward this goal, this talk presents a multi-layered approach that automatically and efficiently plans the sequence of motions the robot needs to execute so that the resulting trajectory is dynamically feasible, avoids collisions with obstacles, and satisfies a given high-level specification. In distinction from traditional approaches in motion planning, the proposed approach can take into account sophisticated high-level specifications given by Finite State Machines, Linear Temporal Logic, STRIPS, Hidden Markov Models, and other planning-domain definition languages. Such expressive models make it possible to specify complex tasks that frequently arise in navigation, manipulation, robotic-assisted surgery, search-and-rescue missions.

Initial validation in physics-based simulations with high-dimensional robotic models demonstrate significant computational speedups over related work and show the ability of the proposed approach to efficiently plan valid trajectories that satisfy complex high-level specifications.

Bio

Erion Plaku is a Postdoctoral Fellow at the Laboratory for Computational Sensing and Robotics at Johns Hopkins University. He received the Ph.D. degree in Computer Science from Rice University in 2008. His research focuses on motion planning and control of cyber-physical systems for human-machine cooperative or fully automatic task performance in complex domains. Some applications include robot navigation, manipulation, haptic exploration, and robotic-assisted surgery. His research interests encompass robotics, hybrid systems, AI, logic, data mining, and large-scale distributed computing.

SANG seminar: On the Internet, "Am I Really not a Dog?''

Friday, November 13, 2009,
12:00 - 01:30pm,
Engineering Building, Room 4201
Michael Sirivianos

Ph.D. student at Duke University

Abstract

Anonymity is one of the main virtues of the Internet. It protects privacy and freedom of speech, but makes it hard to assess the credibility of online users and the content they post. This paper presents FaceTrust, a system that uses online social networks to provide lightweight and attribute-based credentials while preserving a user's anonymity. FaceTrust employs a ``game with a purpose''design to elicit a user's friends' opinions about his self-claimed attributes such as profession or approximate age, and uses attack-resistant trust inference algorithms to estimate the likelihood that an attribute is true. It then provides a credential, which a user can use to corroborate his online assertions. We evaluate FaceTrust using a crawled social network graph as well as a real-world deployment. The results show that our credibility estimations correlate well with the ground truth, even when a large fraction of the social network users are dishonest.

Note:

You can install FaceTrust's Facebook application, called "Am I Really?", at http://apps.facebook.com/am-i-really

Bio

Michael Sirivianos is a Ph.D. student at Duke University. His research interests include network security and peer-to-peer systems. He received a B.S in Electrical and Computer Engineering from the National Technical University of Athens in 2002, and an M.S. in Computer Science from the University of California, San Diego in 2004. More information is available at http://www.cs.duke.edu/~msirivia

CS Seminar: Faculty Research Overview

Wednesday, November 18, 2009,
2:00 PM,
Engineering Building, Room 4201
Prof. Allbeck, Prof. Shehu, Prof. Sousa

Computer Science Department

Abstract

We will have the third monthly seminar reviewing research activities and projects of computer science faculty. Professors Allbeck, Shehu and Sousa will each speak for 15 minutes about their current interests and projects, followed by questions and discussions.

Prof. Allbeck: Places Everyone: Creating an Animated Background of Human Activity Creating virtual scenarios that simulate a human population with typical and varied behaviors can be an overwhelming task. In addition to modeling the environment and characters, tagging the environment with semantic data, and creating motions for the characters, the simulation designer also needs to create character profiles for the population and link these character traits to appropriate behaviors to be performed at appropriate times and places during the simulation. I'll describe an architecture, called CAROSA (Crowds with Aleatoric, Reactive, Opportunistic, and Scheduled Actions), that facilitates the creation of heterogeneous populations for simulations by using Microsoft Outlook®, a Parameterized Action Representation (PAR), and crowd simulator.

Prof. Shehu: Model-based Search for Molecular Structures Search is an important topic in AI. The talk will summarize some robotics-inspired search methods developed in our lab to find biologically-relevant structures of important biomolecules like proteins.

Prof. Sousa: Software issues for Smart Spaces. Smart spaces, aka ambient intelligence, aka ubiquitous computing, enables users to freely interact with multiple computers embedded on furniture, appliances, and vehicles. Or does it? We'll talk about current research issues for software design, end-user programming, human-computer interaction, security and trust, context-awareness, and conflict resolution among multiple users.

Volgenau School Seminar: Opportunistic Spectrum Access in Cognitive Radio Networks

November 18, 2009,
11:30 AM,
Jajodia Auditorium, Engineering Bldg
Dr. Brian Mark

Associate Professor, ECE Department, GMU

Abstract

With the increasing ubiquity of wireless devices, radio spectrum has become a scarce resource. Moreover, recent measurement studies of the radio spectrum have shown that frequency bands allocated to licensed users are highly underutilized. I will give an overview of my recent research on the design and analysis of wireless networks that exploit this underutilized spectrum opportunistically while avoiding harmful interference to the licensed users. The enabling technology for such opportunistic spectrum access is cognitive radio, which is a type of wireless device that can dynamically modify its transmission and reception parameters in response to its radio environment. My research results suggest that opportunistic spectrum access based on cognitive radio technology has the potential to dramatically increase the capacity of wireless networks. Real-world applications and future research directions will also be discussed.

Volgenau School Seminar: The Importance of Models in the Design & Analysis of Computer Systems

November 19, 2009,
3 PM,
Research I, Room 163
Professor Daniel Menasce

Senior Associate Dean, Volgenau School

Abstract

Two important question need to be addressed when designing a computer system: “What functions will it accomplish?” and “How well will it accomplish these functions?” This talk addresses the second question and shows how quantitative models are essential to dealing with that question. A walk through the speaker’s memory lane will highlight a few examples of how such models have been successfully used in the design and analysis of a wide range of environments including: multiprogrammed mainframes, distributed database systems, client-server systems, supercomputers, parallel computing, mass storage systems, web and e-commerce systems, authentication systems, software systems, grid computing, and service oriented architectures. The talk concludes with a discussion on how quantitative models have been used by the speaker in the design of computer systems that are self-optimizing and self-configuring.

Bio:

Daniel Menasce is the Senior Associate Dean at The Volgenau School of Information Technology and Engineering and a Professor of Computer Science at George Mason University. He holds a Ph.D. degree in Computer Science from UCLA, is a Fellow of the ACM, and a recipient of the Computer Measurement Group’s 2001 A.A. Michelson Award for lifetime “outstanding contributions to computer metrics.” Menasce published 200 technical papers and was the chief author of seven books some of which have been translated into Korean, Russian, Portuguese, and Spanish. His research has been funded by DARPA, NASA, NSF, National Geospatial-Intelligence Agency (NGA), Virginia's Center for Innovative Technology, and several private companies. Menascé was the recipient of various prizes including, teaching awards, research awards, and best paper awards. His areas of interest include autonomic computing, e-commerce, performance modeling and analysis, and software performance engineering.

SANG seminar: Security Vulnerabilities in US Voting Machine Systems: A Summary of

Friday, December 04, 2009,
12:00-1:30 pm,
Engineering Building, Room 4201
Micah Sherr

Postdoctoral Researcher- University of Pennsylvania

Abstract

Ensuring reliable elections and increasing the public's trust in the election process are perhaps the two most important responsibilities of our federal, state, and local governments. In most jurisdictions in the United States, elections are managed, configured, and conducted using closed-source and proprietary electronic voting machine software and equipment. Proponents of electronic voting systems argue that these systems are faster, more reliable, more accessible, and more secure than existing voting technologies. This talk discusses the security properties of electronic voting machines, and in particular, highlights numerous discovered vulnerabilities that call into question whether our trust in electronic voting systems is warranted.

In particular, this talk presents the findings from two government-commissioned academic studies of electronic voting machine equipment: the California Top-to-Bottom Review, the first academic review of voting systems in which investigators had access to the systems' source code and developer documentation, and the Ohio EVEREST report, a study of the security and reliability properties of the remaining major voting machine systems that were not included in the California review. In both instances, we found numerous exploitable vulnerabilities in nearly every reviewed system and component. These security flaws enable an attacker to alter or forge precinct results, install corrupt firmware on touchscreen and optical voting hardware, forge paper audit trail entries, and erase electronic log records. In addition to enumerating discovered security flaws, this talk also highlights some of the architectural weaknesses of deployed electronic voting systems, and discusses potential mitigation strategies.

Bio

Micah Sherr is a postdoctoral researcher at the University of Pennsylvania. His academic interests include privacy-preserving technologies, electronic voting security, wiretap systems, and network intrusion detection. He received his PhD in computer and information science from the University of Pennsylvania

CS PhD Dissertation Defense: Computational issues in Long-Term Fairness among Groups of Agents

Monday, December 07, 2009,
10:00AM - 12:00PM,
Engineering Building, 4th flr, Room 4801
Gabriel Catalin Balan

PhD Candidate

Abstract

Fairness within groups is important to a very broad range of problems, from policies for battery-operated soccer robots to distributed traffic control. While no single action may be fair to everyone, it is possible to achieve long-term optimal fairness for everyone through choice of repeated actions I explore the issue of achieving such long-term fairness among multiple agents, and provide a unified view of the problem and solutions to it. The issue of constructing long-term fair policies among multiple agents has not been well studied in the literature. I focus on a particular definition of fairness, called “leximin” fairness, but most of the results apply to other measures as well.

After examination of fairness through an infinite series of repeated actions, I extend analysis in several directions. First, I consider how to achieve as fair a result as possible given a finite series of actions, where the length of the series is not precisely known beforehand but rather is chosen from an unknown or stochastic distribution of time horizons. My solution guarantees the beneficiaries the fairest possible long-term results, minus a bounded worst-case loss due to the game ending unexpectively. I show that finding sequences of actions with optimal worst-case loss is NP-hard, and I propose a family of approximation algorithms.

Second, I examine stateful domains, where one’s choices have side-effects that influence the effects of actions in the future. I introduce a multi-objective genetic algorithm for finding good tradeoff points between beneficiaries utilities and their worst-case losses.

Third, I focus on decision-making processes which have been decentralized in the form of hierarchies. I propose an algorithm based on my stochastic time-horizon solution, and show empirically that an agent hierarchy running that algorithm is able to achieve optimal long-term utilities.

Volgenau School : Non-degree Student Open House

Thursday, December 10, 2009,
6 PM,
Engineering Building, 2nd Floor Lobby area

Abstract

Please visit the open house web page for more information.

Volgenau School of IT&E Graduate Orientation

Wednesday, January 13, 2010,
5:30-8pm,
Innovation Hall Lobby, Fairfax Campus

Abstract

For the fifth year, the Volgenau School of Information Technology and Engineering is welcoming its newly admitted graduate students to a special orientation event.

Although orientation is not mandatory, it is highly recommended that both domestic and international students plan to attend. Essential information regarding university services for graduate students and program information from academic departments will be provided. Also, this is your opportunity to meet your peers, the administrative and academic staff members who will assist you during the pursuit of your graduate course work and degree.

Students admitted for the Spring 2010 semester.

http://volgenau.gmu.edu/new_students/graduate_orientation.php

SANG Seminar: Defending Against Client Compromises in Client-Server Applications

Thursday, January 20, 2010,
2:00-3:00 pm,
Eng 4801
Mike Reiter, PhD UNC Chapel Hill

Abstract

We present new methods for defending against client compromises in two client-server application scenarios. First, we consider online games, in which a client "compromise" reflects the unauthorized manipulation of the game client by the user himself, in order to cheat in the game. To address this threat, we develop a new cheat-detection method with which the server can validate that the messages received from the game client are consistent with the sanctioned client software. Second, we consider a user entering private information to a trusted web server, via a client computer that might be compromised by malware. To address this threat, we leverage trusted computing technology in a novel way to ferry the user's private inputs to the remote server while ensuring that malware cannot capture it.

Speaker's Bio

Michael Reiter is the Lawrence M. Slifkin Distinguished Professor in the Department of Computer Science at the University of North Carolina at Chapel Hill (UNC). He received the B.S. degree in mathematical sciences from UNC in 1989, and the M.S. and Ph.D. degrees in Computer Science from Cornell University in 1991 and 1993, respectively. He joined AT&T Bell Labs in 1993 and became a founding member of AT&T Labs -- Research when NCR and Lucent Technologies (including Bell Labs) were split away from AT&T in 1996. He then returned to Bell Labs in 1998 as Director of Secure Systems Research. In 2001, he joined Carnegie Mellon University as a Professor of Electrical & Computer Engineering and Computer Science, where he was also the founding Technical Director of CyLab. He joined the faculty at UNC in 2007.

Dr. Reiter's research interests include all areas of computer and communications security and distributed computing. He regularly publishes and serves on conference organizing committees in these fields, and has served as program chair for the flagship computer security conferences of the IEEE, the ACM, and the Internet Society. He presently serves on the editorial board of Communications of the ACM, and he has previously served as Editor-in-Chief of ACM Transactions on Information and System Security and on the editorial boards of IEEE Transactions on Software Engineering, IEEE Transactions on Dependable and Secure Computing, and the International Journal of Information Security. He presently serves on the Emerging Technology and Research Advisory Committee for the United States Department of Commerce. Dr. Reiter was named an ACM Fellow in 2008.

SANG Seminar: Energy Management for Time-Critical Energy Harvesting Wireless Sensor Networks

Friday, January 29, 2010,
12:00-1:30PM,
Engineering 4201
Bo Zhang

Abstract

As Cyber-Physical Systems (CPS) evolve they will be increasingly relied on supporting time-critical monitoring and control activities. Further, many CPSs that utilize Wireless Sensor Networking (WSN) technologies will require the use of energy harvesting methods to extend their lifetimes. For this application class, there are currently no effective approaches that balance system lifetime with system performance under both normal and abnormal (emergency) situations. To address this problem, we define a general purpose WSN architecture to support a time-critical CPS system. We then present a set of Harvesting Aware Speed Selection (HASS) algorithms. We use an epoch-based architecture to dynamically adjust radio modulation levels and CPU processing frequencies so that application's end-to-end deadlines are met. Our technique maximizes the minimum energy reserve level for all the nodes in the network, thus ensuring highly resilient performance under emergency or fault-driven situations. We present an optimal centralized algorithm along with a fully distributed solution. Through simulations, we have extensively evaluated our centralized and distributed algorithms against a baseline scheme. Our results show that our algorithms yield significantly higher energy reserve levels than the baseline approach, under both normal and emergency situations.

Bo Zhang is a Ph.D. student in Computer Science Department of George Mason Univesity. His research interests incude Wireless Sensor Networks, Real-Time Embedded Systems and Low-Power Computing. Before pursuing his Ph.D. degree, he received M.S. degree from University of Cincinnati, and B.S. degree from Huazhong University of Science and Technology.

http://cs.gmu.edu/~sqchen/SANG/index.php/SeminarSchedule/Spring2010

Software Engineering Seminar: A Survey on End User Programming Infrastructures for Ubiquitous Computing Environments

Thursday, February 04, 2010,
12:00-1:00PM,
4201 Engineering
Vasilios K. Tzeremes

Abstract

Smart spaces are becoming more and more prominent. The increasing quality of devices, lowering costs, user friendly interfaces, higher network speeds and greater device interconnection are some of the factors that have contributed to this trend. As end users becoming more technologically savvy, they also looking for ways to utilize technology to simplify their everyday life. For example a user might want to create an application that secures his house when he is not present or play his favorite TV show in any of the TVs that he is in front. In this seminar, we will present a collection of end user programming infrastructures that allow end users to create and deploy software applications for their environments. The seminar will begin with a description of major end user programming techniques followed by a collection of end user programming infrastructures for ubiquitous computing. Also for each of the infrastructures, we will provide a detailed description along with benefits and limitations.

Speaker's Bio

Vasilios Tzeremes is a Ph.D. Candidate in Information Technology at George Mason University. Vasilios is working as a Software Engineer in Northern Virginia for the past 9 years. He has developed several software applications for private and government organizations. He specializes in the development of Java Enterprise Edition (Java EE) applications. Vasilios is originally from Greece where he completed his undergraduate studies. In 2004 he completed his Master degree in Information Systems at American University. He is currently working on his dissertation proposal. His dissertation topic focuses on the design and development of a software infrastructure that would enable end user application development for smart spaces.

Student Workshop: Special Behavioral Interviewing Workshop for IT & E Students

Thursday, February 11, 2010,
4:00-5:30pm,
SUB II, Rooms 5 & 6
Ralph Lattanzio, Employer-In-Residence, Exxon Mobil

Abstract

Students will learn and practice the Behavioral Interviewing process.

Engineers Day

Thursday, February 18, 2010,
12:00 to 6:00pm,
Dewberry Hall, Johnson Center

Abstract

This event showcases our faculty and students, departments, research centers and student organizations. Over 30 companies will be in attendance recruiting Volgenau School students and showcasing their companies. The event will include a full schedule of fun activities, presentations and snacks.

SANG Seminar: Placing Intrusion Detection in Context

Thursday, February 18, 2010,
12:00-1:30PM,
Engineering 4201
Michael Collins

Abstract

In network security, traffic analysis is usually seen as a building block towards building a reactive intrusion detection or prevention system - some mechanism that can use traffic data to either inform analysts or actively block hostile traffic before it becomes a significant threat. These goals mean that an IDS is not quite a passive sensor, and not quite a firewall, but somewhere in between. The focus of this talk is on the problem of placing IDS in context with these other defense mechanisms, to do so, I focus on the problems of intelligence, actionability, and payoff.

The problem of intelligence is focused on the distinction between IDS and sensors. We distinguish IDS from simple sensors by their reactive capability --- an IDS somehow informs either operators or networks that some form of defensive action must be taken in response to a current problems. IDS therefore rely on sensors for intelligence, but in comparison to pure sensors, they must make a decision with some potential consequence. This relationship between timing and information gathered means that under specific situations, an IDS may tolerate a high false positive rate in order to provide a rapid response.

The problem of actionability focuses on the relationship between alerts and responses. The majority of attacks taking place on modern networks are effectively harmless --- attackers constantly scan networks for vulnerabilities and automated attack tools try exploiting vulnerabilities on every possible IP address. As a result, simply identifying an "attack" is both insufficient and deceptive - attacks are very easy to find, but the majority of them are effectively harmless. Consequently, an IDS must not only determine whether a system is being attacked, but whether the attack matters.

The problem of payoff focuses on the relationship between IDS and attackers by treating IDS is effectively a design specification. If a rational attacker knows that a particular defense will be applied under certain conditions, then he will act in such a way to avoid triggering the defense. To evaluate the impact of IDS on attackers, we evaluate the relationship between attackers and IDS as a zero-sum game with specific payoff models. By evaluating defensive mechanisms in terms of their payoff, we can potentially unify the problems of training and intelligence into a single mechanism.

Speaker's Bio

Michael Collins is the chief scientist for RedJack, LLC., a Network Security and Data Analysis company located in the Washington D.C. area. Prior to his work at RedJack, Dr. Collins was a member of the technical staff at the CERT/Network Situational Awareness group at Carnegie Mellon University. His primary focus is on network instrumentation and traffic analysis, in particular on the analysis of large datasets and the impact of distributed attacks on Internet infrastructure.

Talk: DNA Technologies: Promises and Computational Challenges

Friday, February 19, 2010,
1:00-2:00pm,
Egineering Rm. 3507
Mihai Pop, Ph.D.

Abstract

During the past few years we have witnessed dramatic advances in DNA sequencing and mapping technologies. These technologies generate data orders of magnitude faster, and at just a fraction of the costs previously possible. As a result, DNA sequencing is rapidly becoming a critical tool in many areas of biology research. At the same time, however, the wealth of data being generated is rapidly challenging the capacity of the computational infrastructure available to researchers. In my talk I will outline several of the applications of the new DNA technologies, and describe recent work in my group aimed at addressing the computational challenges created by the flood of data generated by modern sequencing instruments.

Speaker's Bio

Mihai Pop is an assistant professor in the Department of Computer Science at the University of Maryland, College Park, with a joint appointment in the Center for Bioinformatics and Computational Biology. Prior to his current appointment, he was a Bioinformatics Scientist at the Institute for Genomic Research - a leading genomics institute in Rockville, MD. Dr. Pop's research focuses on computational methods for analyzing high-throughput biological data. Currently, his lab is primarily exploring new approaches for the analysis of metagenomic data.

Grand Seminar: How the Brain Learns Sequences: Towards Self-Programming Robots

Tuesday, February 23, 2010,
12:00 pm,
Engineering 4201
Bruce Swett Intelligent Systems Group of Decisive Analytics Corporation (DAC)

Abstract

The current state of the art in robotics involves replicating selected aspects of human physiology -- i.e., real-time reflexes and postural control; extraction of meaning from sensors; and learning to produce patterns of behavior. The main limitation of robotics lies in the fact that each behavior must be programmed; there is insufficient learning, adaptation, and generalization in machine learning software. Recent research on how the human brain learns can inform the development of novel computational approaches to machine learning. Today I will present neuroimaging research on how the brain learns a novel sequence of movements, as well as my research agenda for translating neuroscience research into machine learning algorithms for self-programming robots.

Speaker's Bio

A Senior Scientist in the Intelligent Systems Group of Decisive Analytics Corporation (DAC), Dr. Swett examines emerging research on how the brain solves complex problems and translates those findings into computational algorithms to address real-world Intel and Department of Defense needs. Dr. Swett has completed post-doctoral training at the National Institutes of Health in both computational modeling of brain processes and experimental neuroimaging of brain functions. At DAC, Dr. Swett has applied his expertise to developing novel computational methods for space situation awareness; automated video exploitation; temporal pattern discovery; machine learning for traffic prediction and vehicle tracking; and probabilistic modeling of human behavior. Dr. Swett has significant experience with high-performance cluster computer programming in the analysis of massive data sets. Dr. Swett is the PI on AF083-038 (Information Fusion and Prediction for Space Situation Awareness -SSA) and AF093-047 (Automated Tools for Adversarial Threat Characterization) with AFRL/RIEA, and DARPA SB082-023 (Query Refinement for Content-Based Video Retrievals). Dr. Swett is DAC’s Technical Lead on artificial intelligence, machine learning and biologically-inspired algorithm research and development.

Grand Seminar: Efficient Motion Planning Algorithms and Applications

Wednesday, February 24, 2010,
12:00 pm,
Engineering 4201
Liangjun Zhang Computing Innovation Fellow Department of Computer Science Stanford University

Abstract

Sampling-based planning algorithms have been widely used to compute collision-free paths for robots. However, their performance can degrade if the free space of a robot has narrow passages. In this talk, I will present efficient sampling methods based on retraction computation and model decomposition. I will describe new geometric algorithm for quantifying the inter-penetration between 3D models, namely generalized penetration depth. These algorithms are used to retract invalid samples to desirable regions and improve the efficiency of planning. In order to deal with many DOFs of human-like robots, robots are decomposed into multiple body components and a constrained coordination scheme is used to plan the lower dimensional sub-problems incrementally. I will demonstrate the efficiency of these planning algorithms and show their applications to virtual prototyping and digital human modeling.

Speaker's Bio

Liangjun Zhang is a Computing Innovation Fellow of the Department of Computer Science at Stanford University. He received his B.S. and M.S. in Computer Science at Zhejiang University, China, and his Ph.D. in Computer Science at University of North Carolina at Chapel Hill in 2009. His research interests include motion planning, geometric computation, robotics and computational structural biology. He received the best paper award at the CAD conference in 2008 and the Linda Dykstra dissertation award of UNC. He also received the 2008 Chinese government award for outstanding Ph.D. students abroad.

Software Engineering Seminar: Self-Assembling Distributed Internet Software

Thursday, February 25, 2010,
12:00-1:00PM,
Engineering 4201
Yuriy Brun

Abstract

Nature uses decentralized mechanisms that can often scale and self-adapt better than human-engineered ones. Certain types of software systems share requirement and resource properties with nature and may benefit from nature-inspired mechanisms. For example, large, highly distributed Internet systems resemble biological bodies with billions of self-contained cells, coordinating to achieve high-level tasks. For such systems, self-management and self-adaptation are critical.

In this talk, I will present the tile style: a nature-inspired architectural style for distributing computation onto large, insecure, public networks, such as the Internet. I will demonstrate how tile-style systems can solve important real-world problems, such as protein folding, image recognition, and resource allocation, while providing guarantees on (1) privacy preservation: tile-style systems preserve the privacy of the algorithm and data, (2) fault and attack tolerance: tile-style systems can tolerate faulty and malicious nodes, and (3) scalability: tile-style systems scale well to leverage the size of the public network to accelerate the computation. In addition to a formal theoretical evaluation, I will discuss an empirical evaluation of a prototype tile-style system deployed on the globally distributed PlanetLab. The analysis shows that problems requiring privacy-preservation can be solved using the tile style orders of magnitude faster than using today's state-of-the-art alternatives.

Speaker's Bio

Yuriy Brun is an NSF CRA postdoctoral Computing Innovation Fellow at the University of Washington. He received his Ph.D. degree in 2008 from the University of Southern California, as an Andrew Viterbi Fellow, and his M.Eng. degree in 2003 from the Massachusetts Institute of Technology. His doctoral research was a finalist in the ACM Doctoral Dissertation Competition in 2008. Brun’s research interests are in the area of engineering self-adaptive and self-managing software systems. His work combines theoretical computer science approaches to modeling nature-inspired algorithms and software engineering approaches to leveraging those algorithms to build systems.

Grand Seminar: Diffusion Tensor Magnetic Resonance Imaging (DT-MRI)

Friday, February 26, 2010,
12:00 pm,
Engineering 4801
Lin-Ching Chang, Assistant Professor Department of Electrical Engineering and Computer Science The Catholic University of America

Abstract

The presentation will begin by talking about the background and basic concepts underlying diffusion tensor magnetic resonance imaging (DT-MRI). Having explained the basic principle, we will then consider how the diffusion tensor is actually estimated from data, what quantitative parameters can be extracted from the tensor, and how the tensor derived quantities can be used in clinical research and applications. The NIH pediatric neuroimaging project (http://www.bic.mni.mcgill.ca/nihpd/info/index.html) will be used as an example to demonstrate how DTI can be used to study normal human brain development. The presentation will pose several problems in DTI processing and analysis, particularly how the artifacts can affect the tensor estimation. Enlightened solutions will be also presented in detail when dealing with artifacts in DTI.

Speaker's Bio

Dr. Lin-Ching Chang is an assistant professor of electrical engineering and computer science at the Catholic University of America, Washington DC, USA. Her research over the past six years has persistently emphasized in the area of magnetic resonance imaging (MRI) processing and analysis. During her career at the National Institutes of Health (NIH), she was working on quantitative image analysis of diffusion tensor magnetic resonance imaging (DT-MRI) data for human brain development. Prior to joining the NIH, Dr. Chang has worked at 3Com Corporation, where she joined and led a number of commercial software projects in telecommunication. Her research interests include software development in medical image analysis, pattern recognition, combinatorial design, information retrial, and telecommunication applications.

SANG Seminar: The Interplay of Dynamic Voltage Scaling and Dynamic Power Management in Real-Time Embedded Applications

Friday, March 05, 2010,
12:00-1:30PM,
Engineering 4201
Vinay Devadas

Abstract

Energy management has become one of the primary goals in the design and development of real-time embedded systems. Dynamic Voltage Scaling (DVS) and Dynamic Power Management (DPM) are two well-known and widely-used energy management techniques. With DVS, the CPU clock frequency and supply voltage can be adjusted to reduce the dynamic energy consumption of the CPU. On the other hand, DPM transitions devices (such as memory and I/O units) to low-power states in an effort to minimize device energy consumption. The research community has extensively investigated both DVS and DPM in disjoint fashion and proposed several effective component-level energy management solutions. Recently, there has been a growing interest in minimizing overall system energy consumption which includes both CPU and device energy. Solutions to this problem must have low run-time overhead and combine both DVS and DPM in a single framework while accounting for their inter-dependencies at the system-level. In this talk, after introducing the basics of DVS and DPM, we will show how their interplay can be characterized in a precise manner for a real-time embedded application. Then, we will present an optimal and efficient algorithm to determine the CPU frequency and device transitioning decisions to minimize the overall system energy.

Speaker's Bio

Vinay Devadas received his MS degree in Computer Science from GMU in 2007. Currently, he is a PhD candidate in Computer Science. His research interests include power-aware computing and real-time systems.

Seminar: Towards High-throughput Phenotyping of Biological Systems: What can bioimage informatics and data mining do for biology

Wednesday, March 17, 2010,
11:00AM,
Engineering 4201
Fuhui Long, PhD, Janelia Farm Research Campus Howard Hughes Medical Institute, VA (Faculty Candidate - Bioengineering)

Abstract

With the advances of fluorescent labeling and bioimaging techniques, biologists can now light up specific proteins in targeted cells or tissues, and acquire 3D+ high-resolution images of an organism both in vitro and in vivo. Such images contain a rich body of phenotype information, including the morphology, anatomy, function, and development of the organisms. In this talk, I will show how bioimage informatics and data mining techniques can help extract phenotype information in an automatic, quantitative, and high-throughput way, and thus turn images into biological knowledge. I will take C. elegans, fruitfly, and human as example systems. Particularly, I will first show how to build digital atlases of model organisms and use them in high-throughput screening of gene expression at the single cell resolution. I will also show how such an analysis can help us gain new insight into the relationships among gene expression, cell fate, and cell lineage in C. elegans. I will then discuss how bioimage informatics techniques can help us reverse engineer a fly’s brain. After that, I will give some examples on human cancer classification and prediction. I will also introduce the high-performance image computing platform, V3D, we have developed that incorporates a variety of bioimage informatics functions including segmentation, registration, modeling, pattern comparison, annotation, and visualization. I will conclude with future research directions.

Speaker's Bio

Dr. Fuhui Long is a staff scientist at Janelia Farm Research Campus, Howard Hughes Medical Institute, working with Dr. Eugene Myers. Her research interest has been focusing on developing efficient bioimage analysis, computer vision, pattern recognition, and data mining techniques to attack challenging problems in neuroscience and molecular biology. Before joining Janelia Farm Research Campus, she was with Lawrence Berkeley National Lab and Duke University Medical Center. Her doctorate is from the Institute of Artificial Intelligence and Robotics from the Xi'an Jiaotong University in 1998. She helped organize several bioimage informatics workshops in the past several years, and is co-organizing the conference “Turning Images to Knowledge: Large-Scale 3D Image Annotation, Management, and Visualization” that will be held this May. Dr. Long is co-supervising postdocs and junior staff scientists in the Myers Lab.

Grand Seminar: Biomechanical Modeling and Simulation of Eye Movements

Tuesday, March 23, 2010,
12:00 pm,
Engineering 4201
Qi Wei Postdoctoral fellow Northwestern University

Abstract

Biomechanical simulation of human eye movements may greatly advance our understanding of the complexities of the oculomotor system and aid in treatment of visuomotor disorders. I will describe the first three dimensional biomechanical model of the orbit that can simulate the dynamics of ocular motility. The model incorporates realistic anatomical and physiological characteristics of the orbital plant, and can implement different types of extraocular pulleys. Various kinds of eye movements such as fixations, smooth pursuits, and saccades can be simulated. It is sufficiently general and adaptable for both scientific research and clinical applications. Several studies were performed to assess the validity and utility of the model. I will also present a template-based approach on reconstructing subject-specific 3D models of the orbit from magnetic resonance imaging (MRI) and an efficient method on estimating longitudinal strains from generalized cylindrical tissues.

Speaker's Bio

Qi Wei is a postdoctoral fellow in the Feinberg School of Medicine at Northwestern University. She received her Ph.D. from Rutgers University in 2010 and her M.Sc. degree from the University of British Columbia in 2004, both in Computer Science. Her research is focused on biomedical imaging, computational modeling and simulation, and eye movement.

Bioengineering Candidate Presentation: Imaging and Quantifying Human Fetal Brain Growth In Utera

Wednesday, March 31, 2010,
10:00am,
4201 Nguyen Engineering Building
Colin Studholme, PhD. Associate Professor, Radiology and Biomedical Imaging, University of California, San Francisco

Abstract

Understanding how the normal human brain develops in utero is a key area of interest in research areas ranging from basic neuroscience to clinical radiology. This talk will describe the work carried out in our group to develop techniques that allow the formation and analysis of high resolution 3D MR images of the unsedated human fetal brain in utero.

Approaches to fetal motion correction and image reconstruction will first be reviewed. These techniques allow the formation of a geometrically correct high resolution 3D MR image by combining multiple motion corrupted clinical 2D acquisitions. These 3D images provide a dramatic new insight into the developing human brain in both normal and abnormal growth, but pose new challenges for automated analysis techniques. Methods for automated atlas based segmentation of both transient and developed tissue structures will be described. These provide accurate tissue maps that form the basis for a range of computational anatomy tools that can be used to reveal the patterns of growth underlying the formation of the human brain. Preliminary results of studies of normal brain growth patterns between the ages of 20 and 27 weeks will be presented in terms of regional tissue volume, deformation tensor morphometry, laminar thickness and surface folding. These methods reveal critical phases of tissue growth and also detect subtle new focal differences related to abnormalities which are of important clinical use. Preliminary results show the promise of these new techniques in providing, for the first time, accurate quantitative maps of early human brain growth in utero.

Speaker's Bio

Dr. Studholme completed a PhD in Medical Physics and Biophysics at the University of London in 1997 in measures of medical image alignment. His postdoctoral work in non-rigid medical image registration was carried out at Yale University where he was awarded a Yale Medical School Postdoctoral Fellowship in Medical Sciences to support his work on incorporating imaging physics into relative MRI distortion correction. He moved as faculty to the University of California San Francisco in 2000 where he is currently Associate Professor of Radiology and Biomedical Imaging, and leads the biomedical image computing group. He is a senior member of the IEEE and has been an NIH funded principal investigator since 2002. He has authored or co-authored over 55 international journal articles on medical image analysis.

Faculty Candidate Seminar: Using (p,n)-grams to Understand Network Traffic

Tuesday, April 06, 2010,
11:00AM,
4201 Nguyen Engineering Building
Abdulrahman Hijazi

Abstract

To better understand the complexity of modern computer networks, we work on finding metrics that can be efficiently measured but that capture patterns in both the headers and payloads of network packets. We have found that many important patterns can be captured using small strings at fixed offsets in packets: (p,n)-grams. We have found that the (p,n)-gram frequency distribution of multi-protocol traffic follows a power law similar to Zipf's law. We also have found that individual protocols have unique (p,n)-gram distributions. In addition to being interesting insights into the structure of modern communications, we use these results to introduce and implement efficient applications for network monitoring, traffic management, and security.

Speaker's Bio

Abdulrahman Hijazi is expected to complete his Ph.D. in Computer Science in April 2010 from Carleton University, Ottawa, Canada. He Completed his Masters degree in Computer Science at Johns Hopkins University and His Bachelors in Computer Engineering at King Fahd University in Dhahra, KSA. For the past four years he has been working with Prof. Anil Somayaji in the Carleton computer Security Lab. Mr. Hijazi worked on several projects in network and Internet security including traffic characterization, protocol identification, and abnormal traffic detection.. He has six publications thus far in peer-reviewed scientific literature and a seventh one submitted to the ACM/IEEE Transactions on Networking

Faculty Candidate Seminar: Steganography and Electronic Voting Security

Wednesday, April 07, 2010,
11:00AM,
4201 Nguyen Engineering Building
Narashimha Karpoor Shashidhar

Abstract

Steganography seeks to address the question "How can two parties communicate secret messages in plain sight without arousing any suspicion?" Traditional Cryptography offers us techniques to transmit encrypted messages. However, the transmission of such messages, no matter how secure, attracts attention! So, how can we cleverly hide secrets within seemingly innocent looking media? I would like to show how Steganography and Information Hiding help accomplish this goal. The second part of my talk is about the Security of Electronic Voting Machines. The electoral process is fraught with challenges - millions of voters to be authenticated, votes to be collected, counted and stored. Now we face an additional new challenge - voting machines with millions of lines of code to be evaluated for security vulnerabilities. I will share my experiences on security and vulnerability analysis conducted on Diebold Optical Scan and Touch Screen systems.

Speaker's Bio

Narasimha Shashidhar is a doctoral candidate in the Computer Science and Engineering Department at the University of Connecticut. He received his Bachelors degree in Electronics and Telecommunication Engineering from the University of Madras, India. He holds a Masters Degree in Computer Science and Engineering from UConn. His research interests are in the fields of Security, Steganography, Cryptography and Electronic Voting. He holds a Graduate Certificate in College Instruction from UConn where he teaches Computer Science and Mathematics.

Grand Seminar: Fast CVT-based Data Visualization Algorithms

Tuesday, April 13, 2010,
12:00 pm,
Engineering 4201
Maria Emelianenko, Ph.D. GMU, Department of Mathematics

Abstract

Efficient data visualization techniques are critical for many scientific applications. Centroidal Voronoi tessellation (CVT) based algorithms offer a convenient vehicle for performing image analysis, segmentation and compression while allowing to optimize retained image quality with respect to a given metric. In experimental science with data counts following Poisson distributions, several CVT-based data tessellation algorithms have been recently developed. Although they surpass their predecessors in robustness and quality of reconstructed data, time consumption remains to be an issue due to heavy utilization of the slowly converging Lloyd iteration. In this talk I will survey several possible approaches to accelerating CVT construction, including a recently developed multilevel optimization based CVT-based binning scheme. Its performance on a set of spectroscopy data, some convergence estimates and possible generalizations will be discussed.

Speaker's Bio

Maria Emelianenko was born in Dubna, Russia. She received an M.S. in Applied Mathematics from Moscow State University and a Ph.D. in Mathematics from Pennsylvania State University in 2005. She spent two years as a Research Associate at Carnegie Mellon's Center for Nonlinear Analysis before joining George Mason University faculty in 2007. She is currently working on problems arising on the interface between mathematics, physics, biology and engineering.

Seminar: Software Development for the iPhone and iPod Touch

Wednesday, April 14, 2010,
1:00-3:00,
Jajodia Auditorium, The Long and Kimmy Nguyen Engineering Builing
Steve Hayman

Abstract

We'll discuss and demonstrate Apple's software development tools for the iPhone and iPod Touch, and show how individuals and institutions are building innovative mobile applications for their organizations and the wider world. We'll highlight and deconstruct some popular applications - both web-based, and native - and show you how to get started building your own with Apple's free developer tools. We'll talk about web development with Dashcode, native application development with Xcode and Interface Builder, and review the integration and deployment options and Apple's developer program. And if the presenter messes up, we may accidentally wind up demonstrating the debugger as well.

Speaker's Bio

Steve Hayman is a National Consulting Engineer with Apple's Education Team based in Toronto, specializing in Apple's developer tools for the iPhone and Macintosh. Prior to Apple, Steve worked with that other Steve at NeXT Computer, where he first fell in love with the combination of powerful object-oriented development tools and a great Unix core; before that he was Network Manager at Indiana University; before that he picked up an M.Math at Waterloo, and before that he had a summer job painting construction equipment. In his spare time he directs Argonotes, the Toronto Argonauts Band, the finest pep band in the Canadian Football League.

SANG Seminar: Time-Bounded Essential Geometrical Localization in MultiHop Wireless Sensor Networks

Thursday, April 15, 2010,
12:00-1:00PM,
Engineering 4201
Wei Cheng

Abstract

Many military and civil applications of a wireless sensor network require the sensors to be aware of their positions. Such a localization problem has been extensively studied in the literature, in terms of both theoretical analysis on the localizability of a sensor network and practical techniques for the actual positioning of sensors. Missing from the existing work, however, is the localization of sensors within a given period of time. In many practical applications of wireless sensor networks, it is crucial to accomplish the localization of sensors within a given time bound because (1) it is on the critical time path - i.e., a sensor has to position itself first before annotating the monitored data with geographical information, and (2) the localization process in general requires consider message exchanges between sensors, making the network more likely to be detected by the enemy. We find that the traditional definition of relative localization - i.e., a process which terminates when all sensors obtain their locations in the same coordinate system is inappropriate for evaluating the actual efficiency of localization in practice, the main reason being that part of the localization process can be seamlessly integrated into subsequent payload transmissions without incurring additional communication overhead. To address this problem, we define a novel problem called essential localization, and present the first study on the essential localizability of a wireless sensor network within a given time bound. http://cs.gmu.edu/~sqchen/SANG/

Speaker's Bio

Dr. Cheng's research interests lie at Mobile computing, Algorithm design and the intersection areas of the two, with primary focus on the problems that arise in the areas of wireless networking and in-network information processing, such as localization, topology control, and channel scheduling. His research has been published in premium networking conferences such as ACM Mobihoc, IEEE Infocom, IEEE ICDCS and journals such as IEEE Transactions on Mobile Computing (TMC).

Joint CS/C4I/SEOR Seminar: Technology: Changing the Game - Impacts of Technological Changes in the Cyber Environment on Software/Systems Engineering Workforce Development

Friday, April 16, 2010,
12:30PM-1:30PM,
Engineering Building, Jajodia auditorium
Kenneth Nidiffer, Ph.D.

Abstract

The world at large is ever more dependent on the advances in the cyber environment. The exponential growth in cyber-reliant systems, technologies, architectures, and capabilities in government and industry organizations represent huge commensurate risks. These risks often occur in the development of software-intensive systems where complex cyber intensive engineering activities continue to suffer from major cost increases, schedules delays, and failures to deliver key capabilities. For numerous reasons, organizations are struggling with the challenges of educating, obtaining and maintaining a best-in-class system and software engineering workforce. A central issue is that although high-level guidance exists, it requires both developers and acquirers to have a workforce with extensive software and systems engineering core competencies, education and experience. Unfortunately organizations are losing their experienced engineering workforce and without capturing this experience, developers and acquirers can misinterpret trade-off alternatives and auger in with disastrous results. This presentation provides an understanding of how technological changes in the cyber environment are causing abrupt changes in software/systems engineering workforce development and some of the solutions that are being proposed.

PhD Defense: Session-Aware RBAC Administration, Delegation, and Enforcement with XACML

Tuesday, April 20, 2010,
10:00-12:00,
Engineering 4201
Min Xu, PhD Candidate

Abstract

An administrative role-based access control (ARBAC) model specifies administrative policies over a role-based access control (RBAC) system, where an administrative permission can change an RBAC policy by updating permissions assigned to roles, or assigning/revoking users to/from roles. Enforcing ARBAC policies over an active access controller while some users are using protected resources may result in conflicts: a policy may be in effect in the RBAC system while being updated by an administrative operation. Towards solving this concurrency problem, this dissertation proposes a session-aware administrative model for RBAC to manage the interactions and potential conflicts between session management and the administrative operations. Based on this model, this dissertation specifies the concurrency requirements of an ARBAC model: (1) revoke an activated role or delete an active session immediately, and (2) delay administrative operations. Consider the eXtensible Access Control Markup Language (XACML) is the de factor language to specify access control policies for Web Services, this dissertation proposes the XACML profile for administrative RBAC (XACML-ARBAC) which is the extension of the XACML-RBAC profile with the proposed session-aware administrative model. One of the advantages of doing so is to use XACML policies to administrate XACML-RBAC policies. This dissertation enhances the XACML evaluation runtime by using a locking mechanism to handle concurrency control issues arising in enforcing the XACML-ARBAC profile. A special administrative policy enforcement point (A-PEP) is developed to compete read-write locks for RBAC and ARBAC policies along with the evaluation engine of an access controller. This dissertation further proposes the XACML-ARBAC profile augmented with role-based delegation, named role-based administration and delegation XACML profile (XACML-ADRBAC). XACML-ADRBAC has two novel properties: scalability--it facilitates delegated permissions to a large number of users with same permission assignment, and flexibility--it allows a delegator to delegate any subsets of permissions assigned to him/her and modify the delegated permission whenever needed. Correspondingly, the proposed XACML-ARBAC enforcement mechanism is also enhanced to enforce the XACML-ADRBAC. To the author's best knowledge, this proposal is the first method to enforce the XACML v3.0 administration and delegation profile proposed by OASIS.

To demonstrate the feasibility and performance of the framework, this dissertation has implemented a prototype to enforce the XACML-ARBAC profile by augmenting Sun Microsystems's XACML reference implementation. Experimental study shows that the system has reconcilable performance characteristics. A copy of this doctoral dissertation is on reserve at the Johnson Center Library.

PhD Defense: The Insider Threat Security Architecture: An Integrated, Inseparable, and Uninterrupted Self-Protection Autonomic Framework

Tuesday, April 20, 2010,
1:00-3:00pm,
Eng 2901
Gihassan Jabbour, IT PhD Candidate

Abstract

The increasing proliferation of globally interconnected complex information systems has elevated the magnitude of attacks and the level of damage that they inflict on such systems. This open environment of intertwined financial, medical, defense, and other systems has attracted hackers to increase their malicious activities to cause harm or to gain unlawful access. However, with the rise of such a problem came the proliferation of a plethora of software tools that claim to solve the problem. A wide variety of software monitoring tools has been deployed to protect against unauthorized access to systems. But, one facet of the problem had been overlooked. Until recently, little or nothing had been done to address the attacks that originate from within the organization. The insider threat did not generally mean much to the organization, specifically to the guardians of its computing infrastructure. In fact, it is the norm to entrust the information system infrastructure to the system and database administrators. But, unfortunately things have changed. The insider, who was always trusted to do what was in the best interest of the organization, is now becoming the one who is, in many cases, harming the organization. News media have reported numerous stories about attacks by insiders and the damage that they caused.

As the insider threat problem started to get recognized, software vendors started to design and deploy new protection systems to address this challenge. However, all of these newly designed approaches have failed, so far, to provide a self-protection mechanism that is innate to the system that is being protected. The premise of this dissertation is based on the notion that providing an uninterruptable autonomic self-protection mechanism that is totally integrated into and inseparable from the computing system that is being protected is critical to ensuring continuous and unconditional protection. This approach to designing system defense mechanisms ensures a solid mitigation to the threat, and an affordable, and assured compliance with system security requirements and government imposed regulations. This dissertation presents solid evidence that demonstrates the seriousness, risk, and malice of security attacks by insiders. Then, it presents the Insider Threat Security Architecture (ITSA) framework and describes its various components. It describes security breach scenarios where privileged users can compromise the computing system that they are entrusted with protecting; then, it shows how the same scenarios can be mitigated under the ITSA framework. The dissertation details the foundational premise that the ITSA framework is built upon. It draws the distinction between the proposed approach and the traditional most common approaches to providing system protection. It emphasizes the unquestionable importance of making the self-protection mechanism as an integral part of the core components of the system that is being protected. A proof-of-concept prototype of the ITSA framework was used by skilled database administrators and security professionals of one of the most security sensitive agencies of the US government. They all found ITSA to be capable of countering the threats that were possible under an equivalent system not protected by ITSA.

A copy of this doctoral dissertation is on reserve at the Johnson Center Library.

PhD Defense: Quality of Service Management of Business Processes in Service Oriented Architecture

Tuesday, April 27, 2010,
2:00pm-4:00pm,
Engineering Bldg., Room2901
Vinod K. Dubey, IT PhD Candidate

Abstract

Service Oriented Architecture (SOA) enables a market of service providers delivering functionally equivalent services at different Quality of Service (QoS) and cost levels. This presents a unique opportunity for consumers to pick and choose services that meet their business and QoS needs. The selected services can be orchestrated in a process flow to optimize the execution of business processes in a cost-effective manner. Given a market of service providers delivering services at different QoS and cost levels, mechanisms should be devised to optimally select services at run-time to support a business process execution so that the selected services together meet end-to-end QoS and cost requirements of the business process and maximize utility for a consumer.

Finding an optimal service selection to support the execution of a business process so that a utility function of a business process is maximized subject to QoS and cost constraints is an NP-hard problem. This dissertation solves this problem by presenting heuristic algorithms that find near-optimal service selections that are very close to the optimal solution while examining a very small portion of the solution space. The proposed methods were evaluated on a very large number of randomly generated business processes as well as experimentally on a proof-of-concept prototype that randomly generates service provider failures and performance degradation.

Computer Science: Awards Dinner

Thursday, May 13, 2010,
6:00pm,
JC Dewberry Hall
Invitation Only

2010: Volgenau Convocation

Thursday, May 13, 2010,
2:00pm,
Patriot Center
Vint Cerf - Vice President, Google Inc

IN CASE OF RAIN students should report directly to the Patriot Center floor, via the lower entrance on the "East" side of the Patriot Center (i.e. loading dock door), to be directed to their seats. There will be no procession in the event of rain.

Abstract

Graduates traditionally assemble at 1:45 p.m. in the Patriot Center parking lot at your department's designated area (indicated by department signs). It is very important that you be at the appropriate location by 1:45 p.m. so we can line up all graduates by degree program for the processional.

At 2:00 p.m., you will process to reserved seating in front of the stage. A brief convocation address will be given by the designated convocation speaker. Following this, the graduates will be called by name to come forth onto the stage to receive their diplomas. Reader cards distributed in the parking lot will be handed to the readers who will read the name of each graduate as they cross the stage. Finally, there will be a reception on the Concourse of the Patriot Center for all in attendance.

The anticipated schedule of events is as follows: 1:45pm Processional Preparation; 2:00pm Processional; 2:30pm Call to Order and Welcome; 2:45pm Commencement Address; 3:00pm Presentation of Diplomas; 4:00pm Reception

There will be no rehearsal, and no tickets are required for this event. You are welcome to bring as many guests as you like. All that you need to do is be sure that you are in the Patriot Center parking lot in the area designated for your department by 1:45 p.m. and that your guests are in the Patriot Center before 2:30 p.m. on the day of the ceremony. Please be sure to wear your cap and gown. Parking for this event will be in Parking Lot A.

We look forward to seeing you and your guests at the The Volgenau School of Information Technology and Engineering Collegial Convocation!

SANG Seminar: NetFence: Preventing Internet Denial of Service from Inside Out

Tuesday, May 25, 2010,
12:00 PM,
Engineering 4801
Prof. Xiaowei Yang

Assistant Professor, CS Dept, Duke University

Abstract

Denial of service (DoS) attacks frequently happen on the Internet, paralyzing Internet services and causing millions of dollars of financial loss. A number of network architectures have been proposed to address the DoS problem, but they all resort to per-host fair queuing to protect legitimate senders when the assumption that receivers can be trusted to expel attack traffic fails. This work presents NetFence, a DoS-resistant network architecture that provably guarantees a sender's fair share of network resources without keeping per-host queues in core network routers and without trusting the receivers. It behaves as effectively as a capability-based system when receivers can identify and bar unwanted traffic. We use a Linux implementation, testbed experiments, ns-2 simulations, and theoretical analysis to show that NetFence is an effective and scalable DoS solution that reduces the amount of state maintained by the bottleneck routers in network core by a few orders of magnitude.

Bio:

Xiaowei Yang is an assistant professor in the Department of Computer Science at Duke University. Before joining Duke, she was an assistant professor in the Department of Computer Science at the University of California at Irvine. She received a PhD in Computer Science from Massachusetts Institute of Technology, and a BE in Electronic Engineering from Tsinghua University. She is a receipt of the NSF CAREER award.

Joint ECE/CS Sang Seminar: Energy Aware Network Operations

Friday, June 11, 2010,
12:00 pm,
Eng 2901
Dr. Puneet Sharma at HP Labs

Abstract

Networking devices today consume a non-trivial amount of energy. This energy consumption is largely independent of the load through the devices. With a strong need to curtail the rising operational costs of IT infrastructure, there is a tremendous opportunity for introducing energy awareness in the design and operation of enterprise and data center networks. In this talk, I will discuss the energy proportionality of today's network devices based on our power profile benchmarking study. I will then present various network energy management techniques and results from two case-studies from a data center network and an enterprise network.

I will also give a brief overview of the networking research at HP Labs.

Speaker's Bio

Puneet Sharma is a Senior Research Scientist at HP Labs where he conducts research on Network Measurement and Monitoring, Wireless Networks, Quality of Service and Data Center Networks. Prior to joining the HP Labs in 1998, he received his PhD in Computer Science from the University of Southern California. He also holds a B.Tech. in Computer Science and Engineering from Indian Institute of Technology (IIT) Delhi. His work on Mobile Collaborative Communities was featured in the New Scientist Magazine. Currently his research focuses on building programmable networks and QoS mechanisms for fabric convergence.

AI Seminar: Alleviating Knowledge Brittleness through Analogical Reasoning and Learning

Tuesday, June 29, 2010,
2:00 PM,
1602 Engineering Building
Matthew Klenk

Postdoctoral Research Associate, Naval Research Laboratory

Abstract

Incomplete knowledge is a source of brittleness for many intelligent agents. When these agents are presented with new situations unanticipated by their designers, they frequently fail to behave intelligently. My research explores the hypothesis that analogy is a crucial mechanism for overcoming this brittleness by applying available knowledge and experience to new situations. Consequently, my research focuses on developing integrated analogical reasoning methods that enable intelligent agents to continually reason and learn about a variety tasks and domains. This presentation consists of two parts. First, I will present two analogical reasoning methods: (1) analogical model formulation for within-domain analogy, (2) domain transfer via analogy for cross-domain analogy. These two methods have been applied to everyday physical reasoning with sketches and physics problem-solving. These tasks require integrating analogy with qualitative and spatial reasoning techniques. The results of empirical evaluations have demonstrated the robust transfer of knowledge from related problems and domains to new situations. In the second part of this talk, I will present the long-term vision of this research, which is to transition computer systems from mere tools to collaborative partners. Building on these methods for within-domain and cross-domain analogy, collaborative agents must perform a range of tasks in a variety of domains drawing on shared experiences with the user. Exploring these ideas through simulated environments requires further integration of planning, execution, and natural communication through language. With more research, collaborative agents could transform our computing experience.

Speaker's Bio

Matthew Klenk is an NRC Postdoctoral Research Associate at the Naval Research Laboratory. He received his PhD in Electrical Engineering and Computer Science from Northwestern University. Matthew's dissertation focuses on analogical reasoning and learning methods for open domains and knowledge-rich tasks. In addition, his research interests include cognitive modeling and integrated AI approaches for achieving human-level intelligence. He received his MS in Computer Science from Northwestern University in December of 2006 and is a member of the Association for the Advancement of Artificial Intelligence (AAAI) and the Cognitive Science Society (CogSci).

Oral Defense of Doctoral Dissertation: Using Image Flow to Analyze Human Gait

Tuesday, June 29, 2010,
9:30-11:30am,
Engineering 3507
Wallace E. Lawson, Jr.

Abstract

Marker-based imaging of human locomotion provides an extremely high level of accuracy, but it is quite intrusive and requires a significant amount of time for both the subject and the gait analyst. The purpose of automated gait analysis is to provide a means to analyze gait from video without the use of markers. Performing this analysis in an automated manner opens up a number of possibilities such as continuous analysis to monitor a course of treatment or to keep watch on the elderly population for changes in gait that might indicate a physical injury or change in mental condition.

There are a number of factors that play into automated gait analysis. Different aspects (or determinants) of gait are active at different parts of the gait cycle. Therefore to provide analysis with respect to all determinants we must have a way of including gait cycle information. There is also the question of how the motion of the limbs can be analyzed. Limbs are constantly self-occluding, and issues such as poor contrast and loose clothing clutter the true motion of the limbs. Actual motion represents the ground truth of how the limb is moving, often times assumed in clinical analysis whereas in automated analysis this is not a given. Loose clothing may obscure actual limb motion, and motion analysis can only provide information about the apparent motion. For these reasons, we see automated gait analysis in some respects as complementary to marker based imaging of gait. It is not possible to have the same level of precision, but the availability and ease of this approach makes it much more applicable to a wider range of scenarios.

In this thesis, we present an approach to automated gait analysis based on the motion of superpixels. We overlay the silhouette of the subject with a regular grid, where each grid cell represents a single superpixel. We overlay an additional superpixel to the top 13 percent of the body, which approximately corresponds to the head region. Human motion analysis is accomplished by analyzing the motion in each of the superpixels. We model the motion of the head using an affine motion model, which can account for a wide variety of valid motions that we can observe in the head (bending at the neck turning in a different direction, etc.) We use a three parameter "twist" motion model on the other regions of the body, which only models the translation and rotation in the superpixel.

Finally, we build a representation of the data using independent components analysis (ICA). ICA provides a compact set of features describing the shape and motion of the body. We use independent components of motion to answer two different questions. The first: can we identify characteristics of the subject (i.e. gender and heel height) given shape and motion information. This is mostly important for identification in a soft biometrics sense. The second: can we identify a person that is walking in a similar manner using ICs. We demonstrate the robustness of the approach by taking it one step further by using ICs of each individual patch to compare the gaits of two individuals, and to give reasons why their gaits are similar and different.

Speaker's Bio

Mr. Lawson holds a BS and MS in Computer Science from George Mason University.

Oral Defense: Improving Arabic Text Processing via Stemming With Application to Text Mining and Web Retrieval

Friday, July 02, 2010,
1:00-3:00,
Engineering 4201
Eiman Tamah Al-Shammari

Abstract

The Internet is loaded every second with massive amounts of open source text data in many languages. The rapid growth of the Internet continues to serve many non-English speaking nations. Arabic is one of the most complex languages, both in its spoken and written forms. It is also the base from which some other languages are derived. Despite the wide usage of the language, technology has been slow in development for Arabic due to the complexity of the written structure of the language. This complexity has created a barrier to text processing advancements that can lead to fully automated Arabic Information Retrieval (IR), Arabic knowledge discovery, and Arabic text mining. The main difficulty facing Arabic language processing is stemming words. Stemming, which is reducing variant words to a single stem, is a text processing step that has a wide degree of variation among languages due to their different morphological structures. Stemming is a recall enhancing approach as well as a feature reduction technique. Existing Arabic stemmers suffer from high stemming error rates. Arabic stemmers blindly stem all the words and perform poorly especially with compound words, nouns and foreign Arabized words. To address the drawbacks of the previous work on Arabic text processing, I propose an alternative methodology that defines rules for stemming words instead of chopping off the letters. The rules are automatically learned by clustering morphologically related words and relying on the input documents only as a content based dictionary. I introduce the “local stem” as a replacement to the current stemming representations and use the concept as a both a stemming algorithm and a query expansion tool. Finally, this research is directed at developing an algorithm for Arabic text processing by overcoming the limitations of previous approaches. The algorithm is tested linguistically using the Paice technique. Furthermore, the algorithm is evaluated on an IR and text mining tasks. Additionally, this research investigates the effect of using full words, stems, and roots of Arabic words as index terms on the accuracy of text mining and the efficiency of the IR of Arabic unstructured documents. The dissertation concludes that the new algorithm outperforms the peer stemmers when adopted in different applications as a stemming and a feature reduction technique. A copy of this doctoral dissertation is on reserve at the Johnson Center Library.

Speaker's Bio

Eiman Tamah Al-Shammari is a PH.D. IT candidate with BS and MS degrees from Kuwait University.

Graduate Students: Non-Degree Open House

Monday, July 12, 2010,
6:00pm-8:00pm,
Fairfax Campus, Nguyen Engineering Building

Abstract

The Volgenau School of IT & Engineering will offer two Non-Degree Open Houses this Summer where prospective students interested in taking graduate coursework for the Fall 2010 term an incredible opportunity to:

* Learn about our graduate programs * Apply as a non-degree student * One-on-one faculty advising available * Obtain an 'on-the-spot' admissions decision, if all eligibility requirements are met*

This is a “one-stop” opportunity to get started on the road towards your graduate education!

A presentation will be given, light refreshments will be served

*Eligibility Disclaimers: 1. Individuals seeking or holding F1 or J1 visas are not eligible for non-degree status, but may apply for any of our degree programs. 2. Individuals must have a minimum cumulative GPA of a 2.5 or better in a related undergraduate degree to be reviewed for admission by the faculty on site. Other applicants will be notified within 3 business days of their admissions decision. 3. Students must submit all required application materials indicated on our Non-Degree Open House requirements checklist. MORE INFORMATION:

RSVPs are required: http://volgenau.ite.gmu.edu/graduateresearch/responseform/

Find Directions/Map: http://coyote.gmu.edu/map Note: We are located in Building #18 on the campus map.

Parking Information/Cost: http://parking.gmu.edu/visitorsregulations.html Note: We are unable to validate parking.

If you have questions about this event, please contact us at itegrad@gmu.edu.

Grand Seminar: Learning Methods for Integrative Cancer Genomics with a Network Perspective

Thursday, July 29, 2010,
11:00AM-12:00PM,
Engineering 4201
Professor Rui Kuang

Abstract

Recently developed high-throughput biotechnologies allow mapping of the complete genome sequence and other molecular features of cancer cells. Finding associations between cancers and various genomic data such as gene expressions and DNA copy numbers can shed light on the molecular mechanisms underlying cancer development and progression. In this talk, I will introduce an alignment-based kernel and two graph-based models to integrate various types of genomic data for identifying biomarkers and classifying tumor samples. Three data integation problems are tackled 1) combining multiple arrayCGH dataset generated from different platforms for DNA copy number analysis; 2) combining differential expression and co-expression among genes to explore the modular structure in genome-wide gene expressions; 3) integrating gene expressions with human protein-protein interaction network as a biological prior knowledge on gene modules. We applied the methods to build predictive models and discover biomarkers from five breast cancer datasets, three lung cancer datasets, two prostate cancer datasets, and three bladder cancer datasets. We show that, while achieving competitive classification results with the data integration models, our methods can also identify more reproducible biomarker than standard statistical measures such as Pearson correlation coefficient and Wilcoxon rank-sum across the multiple microarray datasets. We also performed analysis of biological function and subnetwork enrichment of the identified marker genes and found several enriched biological functions and gene-gene interaction subnetworks of cancer relevance.

Speaker's Bio

Professor Rui Kuang is an assistant professor in the department of computer science and engineering at the University of Minnesota Twin Cities. He specializes in computational biology and machine learning. His research interests are in developing general machine learning approaches for integrative analysis of large-scale genomic and genetic data to understand the molecular characteristics of biological functions and phenotypes. He is particularly interested in designing theoretically principled methods in the categories of graph-based learning methods coupled with optimization techniques, kernel methods, sequence/network alignment methods and bi-clustering/association rule mining methods for a unified analysis of the high-throughput data. Professor Kuang has published over 20 peer-reviewed papers in pretigeous journals and major conferences. He received his PhD in Computer Science from Columbia University.

Dissertation Defense: A Method for Stakeholder-Based Comparative Benchmarking Of Airports

Friday, July 30, 2010,
1:00-3:00pm,
Engineering 4201
Claes Johan David Schaar, PhD IT Candidate

Abstract

Major U.S. airports are critical nodes in the air transportation network, providing the interface between ground and air transportation. Airports are geographic monopolies with multiple stakeholders and operate under profit-neutral financial conditions. Enterprise performance cannot be measured using traditional financial objectives and must instead be evaluated based on the airports’ ability to meet the objectives of all of their stakeholders. Comparative benchmarking is used for evaluating the relative performance of airports. This dissertation describes a systematic method designed for airport benchmarking. The method identifies stakeholders and their goals, selects metrics to reflect the goals, and provides a framework for selection of analytical benchmarking models to handle multi-objective comparisons. Three case study benchmarks of U.S. airports were conducted: (1) A benchmark of the level of domestic passenger air service to U.S. metropolitan areas; (2) a benchmark of the degree of airport capacity utilization; and (3) benchmarks of the level of operational efficiency and investment quality of airports. The implications of the results on government policy, airport improvement funding, and airport management are discussed.

Graduate Students: Non-Degree Open House

Wednesday, August 04, 2010,
6:00pm-8:00pm,
Fairfax Campus, Nguyen Engr Bldg, 2nd Floor Lobby

Abstract

The Volgenau School of IT & Engineering will offer two Non-Degree Open Houses this Summer where prospective students interested in taking graduate coursework for the Fall 2010 term an incredible opportunity to:

* Learn about our graduate programs * Apply as a non-degree student * One-on-one faculty advising available * Obtain an 'on-the-spot' admissions decision, if all eligibility requirements are met*

This is a “one-stop” opportunity to get started on the road towards your graduate education!

A presentation will be given, light refreshments will be served

*Eligibility Disclaimers: 1. Individuals seeking or holding F1 or J1 visas are not eligible for non-degree status, but may apply for any of our degree programs. 2. Individuals must have a minimum cumulative GPA of a 2.5 or better in a related undergraduate degree to be reviewed for admission by the faculty on site. Other applicants will be notified within 3 business days of their admissions decision. 3. Students must submit all required application materials indicated on our Non-Degree Open House requirements checklist. MORE INFORMATION:

RSVPs are required: http://volgenau.ite.gmu.edu/graduateresearch/responseform/

Find Directions/Map: http://coyote.gmu.edu/map Note: We are located in Building #18 on the campus map.

Parking Information/Cost: http://parking.gmu.edu/visitorsregulations.html Note: We are unable to validate parking.

If you have questions about this event, please contact us at itegrad@gmu.edu.

CS Graduate Teaching Assistant: Orientation and Welcome Reception

Thursday, August 26, 2010,
12:00-1:30PM,
ENGR 4201
This event must be attended by all Graduate Teaching Assistants.

CS PhD Program: Welcome and Orientation Reception

Thursday, August 26, 2010,
11:00AM-12:30PM,
ENGR 4201
This event is organized by the CS Department and is mandatory for all incoming CS PhD students.

Machine Learning Seminar: Deep Machine Learning A New Frontier in Machine Intelligence Research

Thursday, September 02, 2010,
12:00 pm,
Engineering 4201
Dr. Itmar Arel, Associate Professor Electrical Engineering & Computer Science, Univ. of Tennessee

Abstract

Mimicking the efficiency and robustness by which the human brain represents information remains a core challenge in artificial intelligence research. Recent neuroscience findings have provided insight into the principles governing information representation in the mammal brain. This discovery motivated the emergence of the subfield of deep machine learning (DML), which focuses on computational models for information representation that exhibit similar characteristics to that of the neocortex. DML offers the ability to effectively process high-dimensional data that may exhibits broad temporal dependencies. This is achieved by employing hierarchical architectures that learn to capture salient spatiotemporal features based on regularities in the observations. In this talk, I will review recent results and chart the future of DML as a field that is bound to have great impact on many areas pertaining to machine learning and intelligent control.

GTA Event: GTA Teaching and Grading Workshop

Tuesday, September 07, 2010,
1:00-3:00pm,
Engineering 4201
Associate Chair Pearl Wang

This event is for all CS Department teaching assistants and faculty. All GTAs are required to attend. Pizza will be served at 12:30pm.

Welcome Reception: For Freshman & New Transfer Students in the BS CS & BS ACS Programs

Wednesday, Septemeber 15, 2010,
12:00-1:15pm,
Research I 163

GRAND Seminar: Spatial Computing: Utilizing spatial principles to optimize distributed computing for enabling the physical science discoveries

Tuesday, October 05, 2010,
Noon to 1PM,
ENGR 4201
Chaowei (Phil) Yang

Associate Professor, Geography, GeoInformation Sciences (GGS), GMU

Abstract

Contemporary physical science studies rely on the effective analyses of geographically dispersed spatial data and simulations of physical phenomena. Single computers and generic high-end computing are not sufficient to process the data for complex physical science analysis and simulations, which can only be successfully supported through distributed computing, best optimized through the application of spatial principles. Spatial computing refers to a computing paradigm that utilizes spatial principles to optimize distributed computers to catalyze advancements in the physical sciences. Spatial principles govern the interaction among different scientific parameters and phenomena across space and time by providing the spatial connections and constraints to drive the progression of the parameters and phenomena. Therefore, spatial computing studies could better position us to leverage spatial principles in simulating physical phenomena and by extension, advance the physical sciences. Using geospatial science as an example, this paper illustrates through three research examples how spatial computing could a) enable data intensive science with efficient data/services search, access, and utilization, b) facilitate physical science studies with enabling high-performance computing (HPC) capabilities, c) empower scientists with multidimensional visualization tools to understand observations and simulations. The research examples demonstrate that spatial computing is of critical importance to design computing methods to catalyze physical science studies with better data access, phenomena simulation, and analytical visualization. We envision that spatial computing will become a core technology that will drive the physical science advancement in the 21st century.

Speaker's Bio

Chaowei (Phil) Yang is the Chief Architect and Technical Lead for NASA Spatial Cloud Computing and Data as a Service (DaaS, hosted by GSFC) and associate professor at George Mason University, where he founded and co-directs (with Dr. Paul Houser, the previous NASA Hydrological Branch Head at GSFC) the NASA/GMU joint Center of Intelligent Spatial Computing for water/energy science (CISC) based on the concept of spatial computing he proposed in 2005 and the hydrological leadership of Dr. Houser. Spatial Computing refers to utilize spatial principles widely exist to arrange, select, and optimize distributed computing to facilitate the advancements of physical sciences, such as Earth and environmental sciences.

His research, education, and service interests include Geospatial Cyberinfrastructure, Distributed GIS, Spatial Computing and Geographic Information Science. He has extensive research and development experience as reflected by his over 50 peer reviewed publications and over $3M research funding in the past decade. He co-edited the Advanced GeoInformation Science book and is writing the book of Network GIS. His research is funded by NASA, UCAR/NSF, FGDC, EPA, NPS, and other agencies/companies with over $3M as PI and He also participate in several large projects total over $10M. He receives numerous national and international awards, such as the US presidential national environment protection stewardship award in 2009.

PhD Defense: Efficient Affine Image Matching for Building and Maintaining 3D Models

Friday, October 08, 2010,
10:00am,
Engineering 4201
Daniel Fleck, Ph.D. CS Candidate

Abstract

3D models of buildings are used in many applications such as location recognition, augmented reality, virtual training and entertainment. Creating models of buildings automatically is a longstanding goal in computer vision research. Many current applications rely on manual creation of models using images and a 3D authoring tool. While more automated approaches exist, they typically are inefficient, require dense imagery, other sensor data, or frequent manual interventions. The focus of this thesis is to automate and increase the efficiency of 3D model creation from image collections. Abstract Matching sets of images to each other is a frequent step in 3D model building. In many applications image matching must be done hundreds or thousands of times. Thus, any in- crease in matching efficiency will be multiplied hundreds or thousands of times when used in these applications. This dissertation presents a new image matching method that achieves greater efficiency by using the fact that images taken from similar viewing angles are approximately related by an affine transformation. An affine transformation models translation, rotation and non-isotropic scaling between image pairs. When images are related by an affine transformation ratios of areas of corresponding shapes are invariant. The method uses this invariant to fit an affine transformation model to a set of putative matches and detect incorrect matches. Methods assuming global and local affine transformation models were created. The first assumes a single global affine transformation between each image pairs. The second method imposes a structure on the feature points to cluster features in a local region. The method then fits different affine models to each cluster. Both methods were evaluated using sets of synthetic matches with varying percentages of incorrect matches, localization error and rotation. Additionally, the methods were applied to a large publicly available image database and the results were compared to several recent model fitting methods. The results show the best affine method using local regions maintains equivalent accuracy and is consistently more efficient than current state of the art methods. When creating and using 3D models, it is often important to predict if images taken from specific locations will match existing images in the model. Image matching prediction is used to evaluate image sets for vision-based location recognition and augmented reality applications. This dissertation presents a new way to predict if images will match by measuring affine distortion. Distortion is measured by projecting features into a second image and computing the affine transformation between the corresponding feature regions. Feature distortion is computed from the skew, stretch and shear of the transformed region. Using the distortion measure for all features in an image pair, a distortion vector is created describing the image pair. Using the distortion vectors and the actual number of matches, a classifier is trained to predict the confidence that images will match. Results are presented that compare this method to other published approaches. The results demonstrate the affine distortion-based classifier predicts matching confidence more accurately than other published techniques. The classifier is also used to create a spatial model of locations around a building. The spatial model shows the confidence that a new image taken from a specific location and pose will match an existing set of images. Using this model, location recognition applications can determine how well they will work throughout the scene. The approach presented uses the classifier described above and more realistic location sampling to create a spatial map that is more accurate than other published approaches. Additionally, as part of this goal, the minimum set of images needed to cover the space around the building is computed. The approach uses structure from motion to create 3D information about the scene. Synthetic cameras are then created using approximate locations and directions from which people commonly take pictures. The affine distortion-based classifier is applied to compute the confidence that images from the synthetic cameras will match the existing set of images. Results are presented on a spatial map showing the confidence that new images captured at specific locations and poses will match the existing image set. Additionally, the minimal set of images needed to maintain the matching coverage is computed using a greedy set cover algorithm. The minimal set can be used to increase efficiency in applications that need to match new images to an existing set of images (e.g. location recognition, augmented reality and 3D modeling applications). Finally, a process is presented to validate the 3D information computed using structure from motion. Validation ensures that the data is precise and accurate enough to provide a realistic 3D model of the scene structure. Results from the process show that the Bundler structure from motion software generates 3D information accurately enough to calculate distortion and generate the spatial coverage map.

Grand Seminar: Vision Based Localization for Robots: metric, topological and semantic mapping

Tuesday, October 12, 2010,
12:00 pm,
Eng 4201
Ana C. Murillo, PhD University of Zaragoza, Spain

Abstract

In the last years, vision sensors have become widespread in many development areas, including robotics, due to several advantages they have with regard to other sensors. In this seminar, we will see some computer vision methods applied in robotics to solve one of the basic issues for any autonomous mobile device: to localize itself and to localize and recognize elements around it. Vision based localization can be solved at various semantic and abstraction levels. In particular, we will see examples of metric localization, using multi-view geometric constraints, topological localization/place recognition approaches and recognition/localization of objects of interest around the robot. Some of this examples, will be presented with experiments run with catadioptric vision systems, very popular in robotics fields, which provide wide field of view to the systems thanks to a curved mirror placed in front of the camera.

Speaker's Bio

Ana C. Murillo is an assistant professor at the Informatic and Engineering Systems Department, a researcher at the Robotics, Perception and Real Time Group, University of Zaragoza, Spain.

PhD Defense: Applications of Logic Coverage Criteria and Logic Mutation to Software Testing.

Monday, October 18, 2010,
10:00-12:00,
Engineering 4201
Garrett K. Kaminski, BA UVA, MS GMU

Abstract

Logic is an important component of software. Thus, software logic testing has enjoyed significant research over a period of decades, with renewed interest in the last several years. One approach to detecting logic faults is to create and execute tests that satisfy logic coverage criteria. Another approach to detecting faults is to perform mutation analysis and then find tests that distinguish the original program from each mutant. The fundamental contribution of this dissertation is the development of a new logic coverage criterion and a new logic mutation approach to improve testing in the context of logic expressions in normal form, logic expressions in general form and entire programs. In particular, testing approaches based on current logic coverage criteria and current mutation approaches share the same drawback of not guaranteeing detection of certain logic faults (even when all non-equivalent mutants are killed) and/or are costly in terms of the number of tests required. This dissertation further develops the body of knowledge in logic coverage criteria and logic mutation testing to address these problems. I show that a new logic coverage criterion can guarantee detecting the same logic faults as current criteria with fewer test cases. I also show that a new logic mutation approach can decrease the number of logic mutants generated while increasing logic fault detection capability. By doing so, a strong theoretical and empirical duality is established between the new logic coverage criterion and the new logic mutation approach.

Grand Seminar: The Role of Data Mining and Statistics in the Financial Service Industry

Tuesday, October 26, 2010,
12:00 -1:00 pm,
Eng 4201
Antonello Loddo, Manager of Statistical Analysis at Capital One

Abstract

The role of data mining and statistical analysis in the financial service industry has changed dramatically due to the increasingly large amount of data that companies are able to process. As a consequence, countless opportunities for data professionals, both inside financial corporations and in analytics companies servicing the sector, have been created. At Capital One, data analysts play important roles in improving tools and processes, and influencing decision-making. We will examine the main problems facing data analysts in financial companies, how they are tackling them, and their impact on strategic and tactical decisions.

Speaker's Bio

Antonello Loddo is a statistical analysis manager at Capital One Financial, where he works on improving tools, processes and best practices for predictive modeling and business decisions. Prior to joining Capital One in 2006, Antonello completed his PhD in statistics at the University of Missouri, Columbia, where he worked as a statistical consultant for the University of Missouri Social Science Statistics Center. Antonello grew up in Cagliari, Italy, where completed his undergraduate studies in economics and worked as econometrician for the transportation research center CRIMM. His interests include Bayesian methods, statistical simulation, machine learning, and model selection.

Grand Seminar: Efficient solutions for large scale learning: applications in speaker recognition and geostatistics

Tuesday, November 09, 2010,
12:00 -1:00 pm,
Engineering 4201
Balaji Vasan, PhD candidate Department of Computer Science University of Maryland.

Abstract

With the ease of data collection, the amount of data available for learning has increased by several folds. This requires any learning technique being used to scale well to large data with many attributes/features. This talk will focus on new machine learning solutions that address the scalability. The applications considered include weather data modeling, speaker recognition and computer vision.

The talk will be divided into two parts; in the first half, the focus will be on Gaussian process regression (GPR). Gaussian process regression is a non-parametric learning technique that has been proven to be robust, but is hindered by its high computational cost. Acceleration of GPR on graphical processors and iterative Krylov solvers will be presented and the framework would be extended to an efficient geostatistical kriging for weather data interpolation. The second half of the talk will focus on a new partial least squares (PLS) regression framework for speaker recognition. PLS has already been applied in several computer vision problems, its extension to speaker recognition in a “supervector” space will be discussed. The new framework has been accelerated on graphical processors, as well.

Speaker's Bio

Balaji Vasan Srinivasan is a PhD candidate in the Department of Computer Science at the University of Maryland, College Park. He works with Prof. Ramani Duraiswami and is also associated with the Perceptual Interfaces and Reality Lab and University of Maryland Institute for Advanced Computer Studies (UMIACS). His current research broadly focuses on developing scalable machine learning solutions for problems in various domains including speaker recognition, weather modeling and computer vision. He completed his Masters in Electrical Engineering in Dec 2008 from University of Maryland with a focus on signal processing and Bachelors in Electrical Engineering in July 2006 from College of Engineering, Anna University, Chennai with a focus on power electronics.

Grand Seminar: Advancing Robotics Through Simulation and Modeling

Tuesday, November 2nd, 2010,
12:00 -1:00 pm,
Eng 4201
Evan Drumwright Assistant Professor Dept. of Computer Science George Washington University

Abstract

Modeling the dynamics of robots is a necessary part of robotics. Roboticists use dynamics models for testing robot code, for optimizing robot controllers, and for learning to perform tasks offline, among other purposes. In the broader picture, these same technologies enable animated movies, more realistic computer games, virtual prototyping, and numerous additional applications. While the science behind such dynamic models (i.e., Newton's laws of motion) is relatively simple, simulating models interacting with contact has proven to be surprisingly difficult. I'll explain why that is the case and describe our solutions to the significant problems that we encountered. I will show how our simulator, Moby, is able to simulate robotic scenarios that other systems cannot handle. The presentation will include many videos, including numerous computer animations simulated using Moby.

Speaker's Bio

Evan Drumwright is an Assistant Professor of Computer Science at The George Washington University in Washington, D.C. Dr. Drumwright completed his Ph. D. in Robotics at the University of Southern California in 2007. His research interests are in building software to control humanoid and manipulator robots and in dynamic robotic simulation. He collaborates with Honda Research Institute in Mountain View, CA and with Willow Garage in Menlo Park, CA to create better robot simulations toward the goal of getting their robots to perform occupational tasks. Dr. Drumwright also develops the free multibody dynamics simulator Moby (http://physsim.sourceforge.net), which is targeted to simulating the dynamics of manipulator and humanoid robots.

Grand Seminar: Querying Similar (Tropical Cyclone) Events via Metric Learning on Multivariate Spatial-Temporal Data Sequences

Tuesday, November 30, 2010,
12:00 -1:00 pm,
Engineering 4201
Shen-Shyang Ho, Assistant Research Scientist, CfAR, University of MD

Abstract

In this talk, I will first provide an overview of my projects that utilize computer science research advances for technology development to support hurricane research. In particular, I will briefly discuss two projects, namely (1) hurricane tracking using heterogeneous satellite data sources, and (2) moving objects database technology to support ad-hoc spatio-temporal query and hurricane data analysis.

Then, I will describe our solution for ad-hoc similarity query based on user-defined instance-level constraints for tropical cyclone events, represented by arbitrary length multivariate trajectory data sequences. A critical component for the solution of such a problem is the similarity/metric function to compare the data sequences. Our solution is a novel Longest Common Subsequence (LCSS) parameter learning approach driven by nonlinear dimensionality reduction and distance metric learning. Intuitively, arbitrary length multivariate data sequences are projected into a fixed dimensional manifold for LCSS parameter learning. Similarity search is achieved through consensus among the (similar) instance-level constraints based on ranking orders computed using the LCSS-based similarity measure.

Experimental results using a combination of synthetic and real tropical cyclone event data sequences are presented to demonstrate the feasibility of our parameter learning approach and its robustness to variability in the instance constraints. I will use a similarity query example on real tropical cyclone events from 2000 to 2008 to discuss (i) a problem of scientific interest, and (ii) challenges and issues related to the weather event similarity search and query problem.

Speaker's Bio

Dr. Shen-Shyang Ho received his PhD in Computer Science from George Mason University in 2007 and his Bachelor (Honors) in Science (Mathematics and Computational Science) from the National University of Singapore in 1999. From 2007 to 2010, he was a NASA postdoctoral fellow and a Caltech Postdoctoral Scholar working at the Jet Propulsion Laboratory (JPL) at the California Institute of Technology. His research interests include artificial intelligence, machine learning, pattern recognition, and data mining for streaming data and on mobile devices. Currently, he is a researcher in the Center for Automated Research (CfAR) of the Institute for Advanced Computer Studies (UMIACS) at the University of Maryland. His current research is a collaboration with JPL and University of Florida, Gainesville, and is funded by NASA.

PhD Defense: A Methodology for Making Early Comparative Architecture Performance Evaluations

Friday, December 03, 2010,
9:00-11:00 am,
Eng 4201
Gerald S. Doyle, PhD CS Candidate

Abstract

Complex and expensive systems’ development suffers from a lack of method for making good architecture-selection decisions early in the development process. Failure to make a good architecture-selection decision increases the risk that a development effort will not meet cost, performance and schedule goals. This research provides a method to mitigate that risk based on the idea that a development can be characterized as the management of uncertainties in a probabilistic experiment. The method developed shows how to estimate the probability that an arbitrary implementation of one architecture will perform better than an arbitrary implementation of an alternate architecture. The analysis technique presented acknowledges that many implementation uncertainties exist at architecture-selection time and identifies steps that can be used to characterize these uncertainties. The process by which uncertainty descriptions are combined into architectural performance descriptions is presented. Once all alternative architecture performance descriptions are developed relative architecture performance comparisons can be made. After the analysis technique is described, three examples are considered. The first example is a simple three tier web-enabled database application. This small web application is used to illustrate the analysis method and demonstrate some methods for characterizing uncertainties. The next two examples are more complex. These examples expose a broader set of uncertainties and show how to handle cases where large numbers of uncertainties exist. Sections on validation of results follow. The paper concludes with a list of future research opportunities in this area.

Speaker's Bio

Gerald S. Doyle, Bachelor of Science, United States Military Academy, 1973 Master of Science, Naval Postgraduate School, 1980, Master of Science, George Mason University, 2000

SANG Seminar: Challenges in Distributed Energy Adaptive Computing

Friday, December 03, 2010,
12:00-1:30PM,
Engineering 4201
Dr. Krishna Kant (NSF & GMU)

Abstract

Fueled by burgeoning online services, power and thermal issues are becoming a substantial issue in terms of cost and environmental impact both on the server (or data center) side and client side. In this talk,we shall motivate an approach that puts energy, power, thermal and sustainability issues at the heart of distributed computing, and strives to dynamically optimize energy consumption based on demands and supply limitations. The talk shall lay out significant challenges in realizing the vision and show some results on how a coordinated power management at multiple levels can help in a graceful QoS degradation in order to adapt to the given energy budgets.

Speakers Bio

Dr. Krishna Kant is currently a visiting research professor at Center for Secure Information Systems at George Mason University, and also serving as a program director in the CISE/CNS division of the National Science Foundation. His current areas of research include power/thermal issues in data centers, robustness in the domain name service, and cloud computing security. He received his Ph.D. degree in Computer Science from University of Texas at Dallas in 1981 and has since held several positions in academia and industry.

SWE Seminar: Visual Language for Software Product Lines in Team Computing & An Inference Network Model for Data Abstraction in a First Response Context

Tuesday, December 07, 2010,
12:00 -1:00 pm,
Eng 4201
Vasilios Tzeremes, IT PhD Candidate, GMU & Salman Salloum Visiting Scholar

Abstract

Visual Language for Software Product Lines in Team Computing: Team Computing (TeC) is a generic end user programming framework that enables users to design and deploy software systems for their environments. TeC follows a visual programming approach. Users drag and drop software components and connect them together to achieve a goal without the need of extensive programming. Software can be deployed at runtime, across spaces, securely without the need of any interruptions. Software Product Lines (SPL) concepts can be used to enhance TeC by allowing end users to design generic software applications that can share with other users. End users will be able to design teams with different features that can be customized at runtime by others. This will increase team quality and simplify the use of complex teams. Currently there are a number of languages used to design and generate SPL members. In this seminar we will present why we believe a visual language may be a better fit for representing SPL in team computing. Moreover and we will discuss some of the challenges that come with it like how do we present SPL primitives in visual languages and how do we support end users in configuring SPL members.

An Inference Network Model for Data Abstraction in a First Response Context: First response workers collaborate in gathering knowledge of the situation on the ground during rescue and evacuation missions. While voice broadcasting is used today for small teams, it does not scale for larger teams or wider deployment areas. However, first responders must pay attention to their mission, and presenting them with all the detailed information gathered from a wide area is unhelpful and distracting. This work proposes a reliable model for data abstraction in a fire response team, taking into consideration that this data is both time-sensitive and space-sensitive, and it may be sensed by different sensors in the area or observed by first responders themselves. The proposed model builds on Inference Networks.

Speaker's Bio

Vasilios Tzeremes is a Ph.D. candidate in Information Technology at George Mason University. He is working as a software engineer in Northern Virginia for the past 9 years. He has developed several software applications for private and government organizations. Vasilios is originally from Greece where he completed his undergraduate studies. In 2004 he completed his master degree in Information Systems at American University. He is currently working on his dissertation proposal. His dissertation topic focuses on supporting end users creating software product lines.

Salman Salloum has an Engineering Diploma (2001-2006) from the Department of Software Engineering and Information Systems, Faculty of Information Technology, Damascus University. He has achieved a remarkable graduate project on developing Reusable Learning Objects, and Developing a Learning Object Repository Supplied by Retrieval and Browsing Capabilities. He is preparing a Master Degree at the same department and focusing his research on the Reusability of Learning Objects. He was leading the eLearning team at ePedia-SY Company (2008-2010). Now, he is an exchange visitor at the Department of Computer Science, George Mason University for Fall 2010 semester.

UNDERGRADUATE: Faculty-Student Winter Mixer

Wednesday, December 08, 2010,
12:00-1:00,
ENGR 4201
Please Join Us!

All Computer Science Undergraduate students, GTAs, UTAs and GRAs are invited to the Computer Science Departments first annual Faculty-Student Winter Mixer. Refreshments will be served.

Oral Defense of Doctoral Dissertation: HYBRID FILTERING IN SEMANTIC QUERY PROCESSING

Friday, December 10, 2010,
10:00am,
Eng 4201
Hanjo Jeong, PhD IT Candidate

Abstract

This dissertation presents a hybrid filtering method and a case-based reasoning framework for enhancing the effectiveness of Web search. Web search may not reflect user needs, intent, context, and preferences, because today’s keyword-based search is lacking semantic information to capture the user’s content and intent in posing the search query. Also, many users have difficulty in representing such intent and preferences in posing a semantic query due to lack of domain knowledge and different schema used by data providers. This dissertation introduces a hybrid filtering method, query-to-query hybrid filtering, which combines semantic content-based filtering with collaborative filtering to refine user queries based not only on an active user’s search history, but also on other users’ search histories. Thus, previous search experience not only of an active user, but also of the other users is used to assist the active user in formulating a query. In addition, a case-based reasoning framework with Semantic Web technologies is introduced to systematically/semantically manage and reuse user search histories for refinement. Finally, ontologies are used for the hybrid filtering to mine preferable content patterns based on semantic match rather than just a keyword match. Validation of the query-to-query hybrid filtering method is performed on the GroupLens movie data sets.

A copy of the doctoral dissertation is on reserve in the Johnson Center library.

CS Phd Defense: Multi-level Sandboxing Techniques for Execution-based Stealthy Malware Detection

Tuesday, December 14, 2010,
10:00-12:00,
Engineering 4201
Lei Liu, PhD Candidate

Abstract

These days all kinds of malware are pervasive on the Internet. Compared to their ancestors that were commonly used for vandalism or demonstration of skills, modern malware, such as Bots, are driven by the underground economics. Often consisting of hundreds to thousands of bots, botnets are one of the most serious threats on the Internet, responsible for various attacks, such as spamming and distributed denial of service (DDoS). As web browsers are the main interface for the majority of Internet users to surf the Internet today, many of such stealthy malware seek to invade via web browsers in the form of browser helper objects (BHO) and browser toolbars.

To defend against Internet malware, existing schemes mainly rely on either signature-based or anomaly-based detection approaches. Signature-based detection is effective for known malware if the malware signature has been generated. However, the effectiveness of signature-based schemes is challenged by polymorphism, metamorphism, obfuscation, encryption, and other techniques. Moreover, signature-based schemes do not work for zero-day (or unknown) malware. On the other hand, anomaly-based detection schemes seek to detect behavior patterns that do not conform to the established normal patterns. Anomaly-based detection schemes do not require malware signatures. However, modern computer software and systems are often complicated, building and analyzing a comprehensive behavior model is time consuming and even impractical.

To overcome these challenges, we propose a novel execution-based approach for stealthy malware detection. In order to facilitate such run-time detection, we aim to design and implement multi-level sandboxing techniques to create controlled running environments to execute testing programs so that their behaviors can be closely observed and analyzed.

First, we leverage virtual machines for OS-level sandboxing to detect bots on individual hosts. By cloning the host image to a virtual machine and screening user input on the virtual machine, the detection noise is significantly reduced. We find that a typical bot exhibits three invariant features along its onset: (1) the startup of a bot is automatic without requiring any user actions; (2) a bot must establish a command and control channel with its botmaster; and (3) a bot will perform local or remote attacks sooner or later. These invariants indicate three indispensable phases (startup, preparation, and attack) for a bot attack. Thus, we propose BotTracer to detect these three phases with the assistance of OS-level sandboxing techniques. To validate BotTracer, we implement a prototype of BotTracer based on VMware. The results show that BotTracer can successfully detect all the bots in the experiments. However, BotTracer may slightly degrade the user performance.

Second, to overcome the limitations of OS-level sandboxes, we build Malyzer based on process-level sandboxes for malware detection. The key of Malyzer is to defeat malware anti-detection mechanisms at startup and runtime so that malware behaviors during execution can be accurately captured and distinguished. For analysis, Malyzer always starts a copy, referred to as a shadow process, of any suspicious process in the process-level sandbox by defeating all startup anti-detection mechanisms employed in the suspicious process. To defeat internal runtime anti-detection attempts, Malyzer further makes this shadow process mutually invisible to the original suspicious process. To defeat external anti-detection attempts, Malyzer makes as if the shadow process runs on a different machine to the outside. Since ultimately malware will conduct local information harvesting or dispersion, Malyzer constantly monitors the shadow process's behaviors and adopts a hybrid scheme for its behavior analysis. In our experiments, Malyzer can accurately detect all malware samples that employ various anti-detection techniques.

Lastly, to detect and contain malicious browser plugins, we develop sePlugin with intra-process sandboxing techniques. With an intra-process sandbox, only plugins are closely monitored for misbehavior detection without confining the entire process. This further reduces the detection overhead while maintaining transparency to end-users. Based on intra-process sandboxing techniques, we build sePlugin to enhance the security of a browser by enforcing security policies on plugins' accessing requests to the browser's internal objects and external system-level resources, such as file systems and network interfaces. sePlugin deals with both native and .NET-based plugins and its unique design renders it possible to work with commodity web browsers without requiring any modifications to the legacy browser architecture or plugin code. We implement sePlugin in Windows XP and IE8.