Banner
Seminars and Events
2011 Events

Elsewhere

Qualifying Exams: Spring 2011 session of qualifying exams for the current PhD students will be held on January 10-14, 2011

Monday, January 10, 2011


The exams' schedule and details will be mailed to the test takers by Wed. 12/08/2010.

Foundation Test-Out Exam: INFS 501

Tuesday, January 18, 2011
2:00pm
Engineering 4201
Registration is required.

Abstract

Please email your intent to csadmin@cs.gmu.edu. Include your name, G number and the exam(s) you wish to register for. A photo ID must be presented on the day of the exam. Each exam will be one hour in length.

It is important to note that you will be permitted to take this exam one time only. Failure to pass the exam will mean that you MUST take the foundation class before enrolling in any core curriculum course.

o

Foundation Test-Out Exams: INFS 501 & INFS 515

Tuesday, January 18, 2011
ENGR 4201

INFS 501 2:00pm-3:00pm
INFS 515 3:30pm-4:30pm
REGISTRATION IS REQUIRED

Abstract

Please email your intent to csadmin@cs.gmu.edu. Include your name, G number and the exam(s) you wish to register for. A photo ID must be presented on the day of the exam. Each exam will be one hour in length.

It is important to note that you will be permitted to take this exam one time only. Failure to pass the exam will mean that you MUST take the foundation classes before enrolling in any core curriculum course.

Foundation Test-Out Exam: INFS 519

Wednesday, January 19, 2011
2:00pm
Engineering 4201
Registration is required.

Abstract

Please email your intent to csadmin@cs.gmu.edu. Include your name, G number and the exam(s) you wish to register for. A photo ID must be presented on the day of the exam. Each exam will be one hour in length.

It is important to note that you will be permitted to take this exam one time only. Failure to pass the exam will mean that you MUST take the foundation classes before enrolling in any core curriculum course.

Foundation Test-Out Exams: INFS 519 & SWE 510

Wednesday, January 19, 2011
ENGR 4201

INFS 519 2:00pm-3:00pm
SWE 510 3:30pm-4:30pm
REGISTRATION IS REQUIRED

Abstract

Please email your intent to csadmin@cs.gmu.edu. Include your name, G number and the exam(s) you wish to register for. A photo ID must be presented on the day of the exam. Each exam will be one hour in length.

It is important to note that you will be permitted to take this exam one time only. Failure to pass the exam will mean that you MUST take the foundation classes before enrolling in any core curriculum course.

GTA Event: Mandatory GTA Orientation Meeting - Pizza will be served.

Wednesday, January 19, 2011
11:00AM - 12:30PM
Eng 4201
Pearl Wang, PHD

Foundation Test-Out Exams: INFS 501 & INFS 515 RESCHEDULED

Thursday, January 20, 2011
ENGR 4201

INFS 501 11:00am-12:00pm
INFS 515 12:30pm-1:30pm
REGISTRATION IS REQUIRED

Abstract

Please email your intent to csadmin@cs.gmu.edu. Include your name, G number and the exam(s) you wish to register for. A photo ID must be presented on the day of the exam. Each exam will be one hour in length.

It is important to note that you will be permitted to take this exam one time only. Failure to pass the exam will mean that you MUST take the foundation classes before enrolling in any core curriculum course.

SWE Seminar: Mathematical Equations as Executable Models of Mechanical Systems

Wednesday, January 26, 2011
3:00 pm
Eng 4201
Walid Taha, PhD. Halmsted Univ. Sweden

Abstract

Cyber-physical systems comprise digital components that directly interact with a physical environment. Specifying the behavior desired of such systems requires analytical modeling of physical phenomena. Similarly, testing them requires simulation of continuous systems. While numerous tools support later stages of developing simulation codes, there is still a large gap between analytical modeling and building running simulators. This gap significantly impedes the ability of scientists and engineers to develop novel cyber-physical systems.

We propose bridging this gap by automating the mapping from analytical models to simulation codes. Focusing on mechanical systems as an important class of physical systems, we study the form of analytical models that arise in this domain, along with the process by which domain experts map them to executable codes. We show that the key steps needed to automate this mapping are 1) a light-weight analysis to partially direct equations, 2) a binding-time analysis, and 3) symbolic differentiation. In addition to producing a prototype modeling environment, we highlight some limitations in the state of the art in tool support of simulation, and suggest ways in which some of these limitations could be overcome.

Speaker's Bio

Prof. Taha is a Professor of Computer Science at Halmstad University in Sweden, and holds and adjunct position at Rice University in Houston, TX. He is credited for developing the idea of multi-stage programming, and is the designer of several systems that develop this idea, including MetaOCaml, ConCoqtion, Java Mint, and the Verilog Preprocessor. He was also involved in the development of several other ideas, including statically typed macros, tag elimination, tagless staged interpreters, event-driven functional reactive programming (E-FRP), the notion of exact software design, and gradual typing. In 2010, Taha's publications had over 1,600 citations, and his h-index was 26. Taha was the principal investigator on a number of research awards and contracts from the National Science Foundation (NSF), Semi-conductor Research Consortium (SRC), and Texas Advanced Technology Program (ATP). He received an NSF CAREER award to develop Java Mint. He founded the ACM Conference on Generative Programming and Component Engineering (GPCE), the IFIP Working Group on Program Generation (WG 2.11), and the Middle Earth Programming Languages Seminar (MEPLS). Taha chaired the 2009 IFIP Working Conference on Domain Specific Languages.

UNDERGRADUATE: Student & Faculty Spring Mixer

Friday, March 04, 2011
12:00pm-1:00pm
Engineering Bld, Room 4201
Please Join Us!

All CS and ACS Undergraduate students and all Computer Science GTAs and UTAs are invited to the Computer Science Department's Faculty-Student Spring Mixer. Refreshments will be served.

SWE Seminar: An Analysis of OO Mutation Operators

Monday, March 07, 2011
12:00 pm
Engineering 4201
Nan Li, CS PhD Student GMU

Abstract

This paper presents results from empirical studies using object-oriented, class-level mutation operators. Class mutation operators modify OO programming language features such as inheritance, polymorphism, dynamic binding and encapsulation. Most previous empirical studies of mutation operators used statement-level operators; this study asked questions about the static and dynamic nature of class-level mutation operators. Results include statistics on the various types of mutants, how many are equivalent, new rules for avoiding creation of equivalent mutants, the difficulty of killing individual mutants, and the difficulty of killing mutants from the various operators. The paper draws conclusions about which mutation operators are more or less useful, leading to recommendations about how future OO mutation systems should be built.

The paper has been accepted for Mutation 2011 in Berlin, Germany. This seminar is a practice for Nan's conference presentation.

Speaker's Bio

Nan Li is a PhD student in Computer Science Department, Volgenau School of Information Technology and Engineering. He received his bachelor’s degree in Software Engineering from Beihang University in China in 2006 and his M.S. degree in Computer Science from Fairleigh Dickinson University in 2008. His current research mainly focuses on mutation testing and model-based testing.

Faculty Candidate Seminar: Seattle: A Peer-to-Peer Platform for Safe Code Execution

Tuesday, March 08, 2011
10:00AM - 11:00PM
Research I, Rm. 163
Justin Cappos, PhD

Abstract

Two of the most significant computing trends of the past five years are peer-to-peer computing and cloud computing. Peer-to-peer systems are powerful in part because they harness under-utilized resources available on end user machines. However, peer-to-peer systems suffer from heterogeneity and a high rate of churn. In contrast, cloud computing allows computation to scale to meet demand via homogeneous virtual environments. However, these resources are often located far from users, are costly, and are restricted by the cloud provider's policies.

The vision of the Seattle project is to provide an infrastructure that gives the best of both worlds. We want to make it practical for arbitrary Internet users to securely participate in a peer-to-peer cloud environment. Seattle has been deployed for two years and has wide spread practical use as a testbed for researchers and educators. The Seattle testbed has been used by 16 classes and is currently supporting the research of dozens of researchers worldwide.

The first part of this talk will give an overview and demo of the Seattle testbed. The second part of the talk will present detailed information about the security architecture for Seattle's sandbox. The security lessons from our sandbox are applicable to similar technologies such as Java.

Speaker's Bio

After obtaining his Ph.D. from the University of Arizona in 2008, Justin Cappos joined the University of Washington as a Post Doc. Justin's research interests generally fall broadly in the area of systems security. He focuses on understanding high-impact, large-scale problems by building and measuring deployed systems. His dissertation work was on Stork, a secure and efficient package manager that has been in use for the past 6 years. Improvements pioneered in Stork have been adopted by most major Linux package managers including APT, YUM, and YaST.

Faculty Candidate Seminar: A Tale of Two Systems: Wireless Privacy and Automotive Security

Tuesday, March 22, 2011
11:00am-12:00pm
Engineering Building, Room 4201
Dr. Damon McCoy

Abstract

Modern products are increasingly designed around connectivity. Once the lone province of computers and phones, today common items such as shoes and cameras, and highly sophisticated devices including implantable medical devices and automobiles, all possess some form of network connectivity. While this transformation has provided a broad range of new capabilities and features, it also opens the door to new security vulnerabilities, as well as privacy threats.

In this talk I will highlight the breadth of these issues, as well as the challenges in addressing them, through two case studies. In the first, I will introduce a taxonomy of privacy threats present in commonly deployed wireless protocols, explain how they manifest practically and finally describe how a privacy-sensitive protocol design can provide equivalent functionality while mitigating those same threats.

In the second part, I will focus on access control threats in the automotive environment. I will describe the attack surface area of modern automobiles, which are largely controlled by internally networked computers and how "transitive connectivity" (both internal and external) creates threats even when no direct communication channel is present. Looking forward, I will discuss the complex challenges in addressing these access control threats.

Candidate's Bio

Damon McCoy is a Computer Innovation Fellow at the University of California, San Diego. He obtained his Ph.D. from the University of Colorado, Boulder. His research includes work on wireless privacy, anonymous communication systems, cyber-physical security, and economics of e-crime. More generally, he is interested in exploring and improving the security and privacy of large-scale systems.

Faculty Candidate Seminar: Enforcement of Privacy Policies: A Logic-Based Approach

Thursday, March 24, 2011
1:00pm-2:00pm
Engineering Building Room 4201
Dr. Deepak Garg

Abstract

Privacy of consumer data is of increasing importance in many areas of commerce including healthcare and finance. Mechanical enforcement of privacy policies is challenging for industry, not only because the often complex and ambiguous policies need to be interpreted by machine, but also because enforcement must be structured to automatically discharge obligations by mining information from system logs. This talk presents recent work on a potential solution to both these problems, using formal logic as a medium to represent and interpret privacy policies and also to structure their enforcement. First, the talk covers an analysis of the U.S. privacy laws HIPAA (for healthcare) and GLBA (for finance), and an ensuing representation of their complex privacy requirements in a new, suitably designed logic. With this representation as a guideline, an algorithm is presented for interactive audit of policy violations. The algorithm is guided by the logical representation of the policy, it is provably sound and it minimizes the need for human input by automatically discharging policy obligations to the extent possible using audit-log data. The talk also covers an ongoing effort to implement and test the algorithm, using standard database interfaces to read audit logs and simulation to create synthetic logs.

Candidate's Bio

Deepak Garg is a post-doctoral researcher in the Cybersecurity Lab (CyLab) at Carnegie Mellon University. He obtained a Ph.D. in Carnegie Mellon's Computer Science Department and an undergraduate degree in Computer Science and Engineering from the Indian Institute of Technology, New Delhi. His research interests are in the areas of computer security and privacy, formal logic and programming languages.

SWE Seminar: Thoughts on Distance Education for the MS-SWE Program

Monday, March 28, 2011
12:00 pm
Eng 4201
Jeff Offutt, Ph.D.

Abstract

GMU has asked the MS-SWE program to offer more classes via distance education. This talk will offer experience-based analysis of the costs and benefits to students, the program, instructors, the department, and the university. The talk will introduce different "flavors" of DE and my base requirements of "students first." The talk will then discuss goals of DE, some of the effects using DE has on the class and the instructor, including specific habits that instructors must adopt or change to use DE effectively. The talk will end with survey results from students in SWE 642, recommendations for using DE effectively, and an answer to the question: "What would it take for the Software Engineering program to make full commitment to distance education?"

Speaker's Bio

Jeff Offutt is Professor of Software Engineering at GMU, where he leads the MS program in Software Engineering.

Grand Seminar: Structure-Activity Relationships and Networks: A Generalized Approach to Exploring Structure-Activity Landscapes

Tuesday, March 29, 2011
11:00AM - 12:00PM
Engineering 4201
Rajarshi Guha

Abstract

Activity cliffs are pairs of molecules that are structurally very similar, yet exhibit very different activities. With this definition one can view molecules and their activities in terms of a landscape - consisting of smooth rolling hills (similar molecules with similar activities) and jagged cliffs. The landscape view provides a framework within which one can analyse structure-activity relationship (SAR) models. The first step is to numerically characterize the landscape and we have devised the Structure Activity Landscape Index (SALI) to do this. This value can be used to construct a network model of the dataset, that allows one to interactively explore activity cliffs of varying degree. While a useful and intuitive visualization of SAR's, the SALI allows us to go further and quantitatively assess the ability of models to encode the SAR's present in a dataset. I will highlight discuss an extension of the SALI, termed SALI curves, to determine how well an SAR model has been able to encode the landscape. I will highlight its generality by applying it to QSAR, docking and CoMFA models. I will also briefly describe the utility of the SALI values to assess the suitability of a given molecular represention for predictive modeling. Finally, I will dicuss more recent work on the predictive model of SALI values, as a way to extend a structure activity landscape as well as prioritize new molecules as part of putative activity cliffs.

Speaker's Bio

Rajarshi Guha is a Research Scientist at the NIH Chemical Genomics Center, working on cheminformatics and bioinformatics problems in high throughput screening. Recently, he has developed the informatics infrastructure for the Trans-NIH RNAi Screening Initiative and is interested in developing strategies to integrate small molecule and RNAi screening data generated using high-content methods. He also holds an adjunct professorship at Indiana University in the School of Informatics and is the Chair-Elect of the ACS Division of Chemical Information.

Oral Defense of Doctoral Dissertation: Simulation-based Stochastic Optimization on Discrete Domains: Integrating Optimal Computing and Response Surfaces

Friday, April 01, 2011
2:00 - 4:00 pm
Eng 2901
Mark W. Brantley

Abstract

Simulation can be a very powerful tool to help decision making in many applications but exploring multiple courses of actions can be time consuming. Numerous ranking & selection (R&S) procedures have been developed to enhance the simulation efficiency of finding the best design. This dissertation explores the potential of further enhancing R&S efficiency by incorporating simulation information from across the domain into a regression metamodel. Under some common conditions in most regression-based approaches, our new method provides approximately optimal rules that determine the design locations to conduct simulation runs and the number of samples allocated to each design location for problems with only one partition. In addition to utilizing concepts from the design of experiments (DOE) literature, it introduces the probability of correct selection (PCS) optimality criterion that underpins our new R&S method to the DOE literature. This dissertation then extends the method by incorporating simulation information from across a partitioned domain into a regression based metamodel. Our new method provides approximately optimal rules for between and within partitions that determine the number of samples allocated to each design location. Numerical experiments demonstrate that our new approaches for one partition domains and for multiple partition domains can dramatically enhance efficiency over existing efficient R&S methods.

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

Faculty Candidate Seminar: Stronger Foundations for Cryptography: Security Against Related-Key Attacks and New Constructions From Lattices

Tuesday, April 05, 2011
10:00am-11:00am
Engineering Building Room 4201
Dr. David Cash

Abstract

In this talk I will discuss two projects that develop new techniques for constructing provably-secure cryptographic algorithms.

In the first part, I will present the first construction of a blockcipher that provably resists ``related-key attacks (RKAs).'' Blockciphers are a ubiquitous tool at the heart of essentially all practical cryptography, and while blockcipher designers had generally agreed that RKA-resistance was necessary, prior to this work there was little evidence to suggest that it was possible at all. Our result gives theoretical validation to this practical security goal.

In the second part, I will present new techniques for building cryptographic algorithms based on lattices. Lattices offer a number of advantages over the traditional number theory used in cryptography, including resistance against quantum attacks and worst-case/average-case equivalence. Using our techniques, we obtain several constructions for tasks that were previously beyond the reach of provably-secure lattice-based cryptography due to the inherently different structure of lattices.

Candidate's Bio

David Cash is a postdoctoral scholar at the University of California, San Diego. He recently completed his PhD in Computer Science at Georgia Tech.

Faculty Candidate Seminar: Exploring game-based learning: How to get students to write code (and like it)

Thursday, April 07, 2011
1:30-2:15 pm
Eng 4201
Dr. Sue Kase

Abstract

Many believe the study of computer science is for stu- dents with exceptional mathematics and technical skills. This course aims to reposition computing in a way that is more attractive to students normally intimidated by formal computer science courses. The hands-on demonstration (instead of lecture) format in a laboratory setting promotes an engagement-driven informal learning environment. The introduction of computer programming concepts in the context of game development adds an element of “fun” to what typically (and kindly) is described as a “dry and technical” area of study. The use of open source soft- ware and online documentation reduces the student’s fi- nancial barriers when taking a course outside their major. Although the course is self-contained within a 15-week semester, potential outcomes include: motivate student learning in areas related to computer science; experimen- tation with other programming languages and computa- tional applications; utilization of new computational skills in students’ major area of study; and transference of pro- cedural and logical thought processes into students’ life experiences. The presentation will briefly discuss this newly developed course and demonstrate example in-class activities.

Faculty Candidate Seminar: Concepts Before Details

Tuesday, April 12, 2011
1:30-2:15 pm
ENGR 4201
Mark Snyder, University of Kansas PhD Student

Abstract

Computer science as a discipline is the study of com- putation, of algorithms, of data. When we prepare students to enter the workforce with a degree in com- puter science, it is this set of working knowledge--not expertise in particular tools--that the degree signifies. We should emphasize the theory and concepts inde- pendent of particular implementations concretely, only then investing our class time in any particular language or tool. In this talk, I briefly discuss what I believe computer science is, as well as some broad rules to follow par- ticularly when teaching new students. I discuss the purpose and feasibility of feedback and assessment in the computer science discipline, and offer an ex- ample of introducing basic object oriented concepts.

Faculty Candidate Seminar: "Lesson Learned: Approaches to Teaching Undergraduate Computer Science"

Thursday, April 14, 2011
1:30-2:15pm
ENGR 4201
Dr. Kinga Dobolyi

Abstract

"One purpose of an undergraduate education is to acquire skills to be applied later in life. When teaching computer science, such skills include not only concepts that we conventionally think of as necessary to impart on students (such as programming languages, tools, and methodologies), but also abstract skills related to problem solving, creativity, and self-management. In this talk I focus on my approach towards educating undergraduate computer science students within the context of passing on such skills, including lessons I have learned from my first year of full-time teaching. Specifically, I discuss the challenges associated with having large class sizes, students of varied background and maturity in the same class, and different attitudes towards introductory programming courses versus more theoretical material. As part of my presentation I will demonstrate the application of many of my teaching methodologies in a short example lecture."

Grand Seminar: What does it take to write efficient proximity algorithms (for robotics, graphics and CAD)?

Friday, April 15, 2011
1:00 PM
Eng 4201
Young J. Kim, Associate Professor Ewha Womans University

Abstract

Proximity query is a process to reason and derive a geometric relationship among, often time-dependent, objects in space. Typical proximity queries include - continuous and discrete collision detection, Euclidean distance computation, Hausdorff distance computation, penetration depth computation, etc. These queries play a vital role in diverse applications such as non-smooth contact dynamics, robot motion planning, physically-based animation and CAD disassembly planning. In this talk, we will focus on our recent research efforts to devise fast - mostly real-time - and efficient proximity algorithms with different objectives using rather simple approaches, thereby working extremely well in practice. Moreover, we also discuss how we can parallelize these calculations utilizing modern hardware platforms such as multi-core CPUs and GPUs. Finally, we demonstrate how to apply these queries to the aforementioned applications, particularly for robot motion planning and contact dynamics.

Speaker's Bio

Young J. Kim is an associate professor of computer science and engineering at Ewha Womans University. He received his PhD in computer science in 2000 from Purdue University. Before joining Ewha, he was a postdoctoral research fellow in the Department of Computer Science at the University of North Carolina at Chapel Hill. His research interests include interactive computer graphics, computer games, robotics, haptics, and geometric modeling. He has published more than 50 papers in leading conferences and journals in these fields. He also received the best paper awards at the ACM Solid Modeling Conference in 2003 and the International CAD Conference in 2008, and the best poster award at the Geometric Modeling and Processing conference in 2006. He was selected as best research faculty of Ewha in 2008, and received the outstanding research cases award from Korean research foundation in 2008.

Grand Seminar: Classification, Clustering and Data Mining of Biological Data

Thursday, April 21, 2011
12:00 pm
Eng 4201
Peter Revesz, Department of Computer Science and Engineering University of Nebraska-Lincoln and Jefferson Science Fellow U.S. Department of State

Abstract

The proliferation of biological databases and the easy access enabled by the Internet is having a beneficial impact on biological sciences and transforming the way research is conducted. There are currently over 1100 molecular biology databases dispersed throughout the Internet. However, very few of them integrate data from multiple sources. To assist in the functional and evolutionary analysis of the abundant number of novel proteins, we introduce the PROFESS (PROtein Function, Evolution, Structure and Sequence) database that integrates data from various biological sources. PROFESS is freely available at http://cse.unl.edu/~profess/. Our database is designed to be versatile and expandable and will not confine analysis to a pre-existing set of data relationships. Using PROFESS, we were able to quantify homologous protein evolution and determine whether bacterial protein structures are subject to random drift after divergence from a common ancestor. After relevant data have been mined, they may be classified or clustered for further analysis. Data classification is usually achieved using machine-learning techniques. However, in many problems the raw data are already classified according to a set of features but need to be reclassified. Data reclassification is usually achieved using data integration methods that require the raw data, which may not be available or sharable because of privacy and legal concerns. We introduce general classification integration} and reclassification methods that create new classes by combining in a flexible way the existing classes without requiring access to the raw data. The flexibility is achieved by representing any linear classification in a constraint database. We also considered temporal data classification where the input is a temporal database that describes measurements over a period of time in history while the predicted class is expected to occur in the future. We experimented with the proposed classification methods on five datasets covering the automobile, meteorological and medical areas and showed significant improvements over existing methods.

Speaker's Bio

Peter Revesz holds a Ph.D. degree in Computer Science from Brown University. He was a postdoctoral fellow at the University of Toronto before joining the University of Nebraska-Lincoln, where he is a professor in the Department of Computer Science and Engineering. His current research interests are bioinformatics, geoinformatics and databases, in particular constraint, genome, spatial and temporal databases, and data mining. He is the author of the textbook Introduction to Databases: From Biological to Spatio-Temporal (Springer, 2010). He held visiting appointments at the IBM T.J. Watson Research Center, INRIA, the University of Hasselt, the Max Planck Institute for Computer Science, the University of Athens, and the U.S. Department of State. He is a recipient of an Alexander von Humboldt, a J. William Fulbright, and a Jefferson Science Fellowship.

CS Phd Defense: Nonparametric Bayesian Models for Unsupervised Learning

Monday, April 25, 2011
11:00 am
Eng 4201
Pu Wang

Abstract

Unsupervised learning is an important topic in machine learning. In particular, clustering is an unsupervised learning problem that arises in a variety of applications for data analysis and mining. However, clustering is an ill-posed problem and, as such, a challenging one: no ground-truth that can be used to validate clustering results is available. Two issues arise as a consequence. Various clustering algorithms embed their own bias resulting from different optimization criteria. As a result, each algorithm may discover different patterns in a given dataset. The second issue concerns the setting of parameters. In clustering, parameter setting controls the characterization of individual clusters, and the total number of clusters in the data. In addition, the high-dimensionality of the data, which is commonly seen in practice, makes the clustering process even more difficult.

Clustering ensembles have been proposed to address the issue of different biases induced by various algorithms. Clustering ensembles combine different clustering results, and can provide solutions that are robust against spurious elements in the data. Bayesian approaches have been applied to clustering to address the parameter tuning and model selection issues. Bayesian methods provide a principled way to address these problems by assuming prior distributions on model parameters. Prior distributions serve as regularizers for modeling parameters, and can help avoid over-fitting. A special kind of Bayesian methods, nonparametric Bayesian approaches, have been proposed to address the key question: "How many clusters?". Nonparametric Bayesian models allow the number of parameters to grow with the number of observations. After observing the data, nonparametric Bayesian models fit the data with finite dimensional parameters.

Although attempts have been made in the literature to address individually the major issues related to clustering, no previous work has addressed them jointly. In my dissertation I introduce a unified framework that addresses all three issues at the same time. I designed a nonparametric Bayesian clustering ensemble (NBCE) approach, which assumes that multiple observed clustering results are generated from an unknown consensus clustering. The underlying distribution is assumed to be a mixture distribution with a nonparametric Bayesian prior. The number of mixture components, i.e., the number of consensus clusters, is learned automatically. By combining the ensemble methodology and nonparametric Bayesian modeling, NBCE addresses both the ill-posed nature and the parameter setting/model selection issues of clustering. Furthermore, NBCE outperforms individual clustering methods, since it can escape local optima by combining multiple clustering results.

I also designed a nonparametric Bayesian co-clustering ensemble (NBCCE) technique.

NBCCE inherits the advantages of NBCE, and in addition it is effective with high dimensional data. As such, NBCCE provides a unified framework to address all the three aforementioned issues.

I have performed extensive evaluation on relational data and protein-molecule interaction data. The empirical evaluation demonstrates the effectiveness of NBCE and NBCCE and their advantages over traditional clustering and co-clustering methods.

Speaker's Bio

Pu Wang, received his Bachelor of Engineering, Mechanics from Beihang University, Beijing, China, 2004 and his Masters of Science, Computer Science from Peking University, Beijing, China, 2007

Grand Seminar: Signal Detection using Text Mining in Large Document Repositories

Tuesday, April 26, 2011
12:00-1:00 pm
Eng 4201
Sithu D. Sudarsan

Abstract

Data mining has traditionally been used with structured data, which is only about 10 - 15 percent of all data in any organization. The remaining 80 - 85 percent data is in the form of free text documents with unstructured narratives. The Center for Devices and Radiological Health at the US Food and Drug Administration receives a large number of regulatory documents related to premarket and postmarket activities for medical devices. One such document is a medical device adverse event report (MDR) that contains multiple narratives related to any adverse event related to a medical device. These MDR narratives need to be analyzed by mining them to identify safety signals. Here, signal refers to either a change in the pattern of occurrence of the adverse event or identification of an unknown adverse event for a given device. The Center receives as many as a few hundred thousand MDRs every year. This repository is growing and of the order of millions of documents running into several terabytes. A text mining framework has been developed by the software research team to identify, and evaluate signals as a research project. The presentation covers the challenges, and solution approach of the framework. Some of the mining results with respect to the MDR repository are also presented.

Speaker's Bio

Sithu D Sudarsan, a Visiting Scientist with the Division of Electrical and Software Engineering in the Office of Science and Engineering Laboratories at the Center for Devices and Radiological Health of US Food and Drug Administration (FDA). In his current role, Sithu is responsible for the Design and Development of Semantic Text Mining Framework with his colleagues. Sithu obtained his Ph. D. from UALR, Little Rock for his thesis titled, “Signal Detection Framework using Semantic Text Mining Techniques”. He received “Outstanding Ph D Graduate for the year 2010” from UALR. Prior to joining FDA he held several positions at leading research laboratories in India. He completed his Bachelor's in Electronics & Communication Engineering, and Mater's in Systems & Information as well as Post Graduate Diploma in Management and Diploma in Production Management.

As an R&D engineer, over the last 20 years he has worked in different fields including Communication, Networks, Information Security, Data Mining and Multi-media. He has received several research grants and has been on several committees that monitor funded R&D projects. He was an active member of IT Standardization Sub Committee (ITSSC) of Ministry of Defence representing M/s Bharat Electronics Limited (BEL), India. He was also member of Bureau of Indian Standards (BIS) committee on Information Security, LTD38 as a nominated representative of BEL. As a life member of IETE, he contributes to the professional development society and held different positions of IETE Noida Chapter during 1998-2001. He has over 50 papers/publications to his credit at international and national levels. He has conducted a number of seminars and workshops in his area of work.

CS/ACM Distinguished Lecture series: The D Programming Language

Tuesday, April 26, 2011
7:00 PM
Research I, Rm. 163
Walter Bright

Abstract

Join us for a discussion led by Walter Bright on the D Programming Language and its programming community.

The D programming language, D is an advanced systems programming language. It is designed to combine the power and high performance of C and C++ with the programmer productivity of today's modern languages such as Ruby and Python. While ts syntax style resembles C/C++, D is multiparadigm language and supports many styles of programming including: imperative, object-oriented and metaprogramming.

Speaker's Bio

Walter Bright graduated from Caltech in 1979 with a degree in mechanical engineering. He worked for Boeing for 3 years on the development of the 757 stabilizer trim system. He then switched to writing software, in particular compilers, and has been writing them ever since.

Grand Seminar: Inferring Non-Observable Object Properties

Thursday, May 26, 2011
11:00 am
Eng 4201
Hedvig Kjellström, Associate Professor Royal Institute of Technology, Stockholm, Sweden

Abstract

The great majority of object analysis methods are based on visual object properties - objects are categorized according to how they appear in images. Visual appearance is measured in terms of image features (e.g., SIFTs) extracted from images or video. However, besides appearance, objects also have many properties that can be of interest, e.g., for a robot who wants to employ them in activities: Temperature, weight, surface softness, and also the functionalities or affordances of the object, i.e., how it is intended to be used. One example, recently addressed in the vision community, are chairs. Chairs can look vastly different, but have one thing in common: they afford sitting. At the Computer Vision and Active Perception Lab at KTH, we study the problem of inferring non-observable object properties in a number of ways. In this presentation I will describe some of this work. These methods can be seen as ways of incorporating contextual information in object detection and recognition. I will also describe our work on contextual human pose estimation: two different approaches to improving the estimation of human pose using object context.

If you would like to meet with the speaker contact kosecka@cs.gmu.edu.

MS Thesis Defense: Variable-Length Fragment Assembly within a Probabilistic Protein Structure Prediction Framework

Thursday, June 16, 2011
10:30am
Eng 4201
Kevin Molloy

Abstract

It is widely accepted that a protein's biological function is highly correlated to the three-dimensional shape the protein assumes under physiological/native conditions. Predicting this three-dimensional structure, known as the native structure, from the protein's amino acid sequence is known as the protein structure prediction problem. This problem is regarded by many to be the of the grand challenges of computational biology.

Fragment-based assembly is a widely used technique in ab-initio structure prediction methods that seek to predict structure from sequence. Essentially, a protein structure is pieced together with configurations of fragments extracted from databases of deposited protein native structures. Fragment length is an important consideration. The shorter the fragment, the more complex the protein conformational space where the native structure resides and the more rugged the energy surface associated with that space. The longer the fragment, the simpler the conformational space and the smoother the energy surface; hence, the higher the risk of missing important regions of space that may lead to the native structure.

In this thesis, we explore the idea of varying the employed fragment lengths to alter the protein conformational space explored during a probabilistic search for the native structure.

Varying fragment lengths allows for manipulating the dimensions of the search space during the process of sampling protein conformations. Essentially, longer fragments are used in early stages of the search to simplify the search space and smooth the energy surface. Shorter fragments are then utilized in later stages to provide visibility to the more complex and realistic conformational space.

This approach is validated on four protein systems of diverse sizes and native topologies. The results show that employing variable-length fragments enhance the sampling of the conformational space for each protein, producing higher-quality native-like structures as compared to using a single fragment length. These promising results lay the foundations for exploring additional research directions in equipping a probabilistic search framework with the ability to make on-the-fly decisions and adaptively change the dimensionality of the conformational space it explores.

Thesis Director Prof. Amarda Shehu

Faculty Candidate Seminar:: Service Integrity Assurance in Large-Scale Cloud Systems

Wednesday, June 22, 2011
11:00am-12:00pm
ENGR 4201
Juan Du

Abstract

Cloud systems have emerged as promising service provisioning platforms for application service providers (ASPs). In open cloud systems consisting of a large number of service providers from different security domains, we can no longer assume all service providers are trustworthy and strive to provide the promised services. Moreover, cloud systems often host long-running applications such as massive data processing, which provides more opportunities for attackers to exploit the system vulnerability and perform strategic attacks. One of the key security concerns for cloud-based data processing is the result integrity, no matter whether private or public data are processed by the cloud system. In this talk, I will present RunTest, a scalable runtime integrity attestation framework to assure the integrity of dataflow processing in large-scale cloud systems. RunTest provides light-weight application-level attestation to dynamically verify the integrity of data processing services and pinpoint malicious service providers. At the end of the talk, I will also talk about my past teaching experience and my teaching philosophy.

Oral Defense of Doctoral Dissertation: Decision-Guided Recommenders with Composite Alternatives

Monday, July 11, 2011
11:00AM - 12:00PM
Eng 1602
Khalid I. Alodhaibi

Abstract

Recommender systems aim to support users in their decision-making process while interacting with large information spaces and recommend items of interest to users based on preferences they have expressed, either explicitly or implicitly. While state-of-the-art recommender systems focus on atomic products or services, this research focuses on developing a framework, models and algorithms for recommending composite services and products based on decision optimization. Composite services are characterized by a set of sub-services, which, in turn, can be composite or atomic and make the recommendation space very large (or infinite, for continuous case).

The proposed framework contains models that allow for fast and easy user preference elicitation that can be captured in a utility function, and provides algorithms for diversifying a recommendation set. Such recommendations will be dynamically defined using database views extended with decision optimization based on mathematical programming. A key challenge addressed in this research is combining the flexibility of diversity ranking functionality with the capabilities of information processing to learn and capture users’ preferences through an iterative learning process.

Oral Defense of Doctoral Dissertation: System-Level Energy Management for Real-Time Systems

Tuesday, July 12, 2011
9:30-11:30AM
Eng 4201
Vinay Devadas

Abstract

Energy management has recently become one of the key dimensions in the design of real-time embedded systems. While early studies focused separately on individual energy management techniques targeting different system components, there is growing interest in system-level energy management frameworks that exploit multiple techniques simultaneously.

A primary objective of this dissertation is the integration of two well-known energy management techniques Dynamic Voltage Scaling (DVS) and Dynamic Power Management (DPM). With DVS, the supply voltage and clock frequency of the processor can be scaled down at run-time, to save CPU energy at the expense of increased task response times. On the other hand, DPM targets reducing the energy consumption of idle off-chip system devices such as disk and memory modules, by transitioning them to their low-power sleep states. While effective system-level energy management mandates the use of both DVS and DPM, their integration poses several challenges. For instance, minimizing device energy requires running the processor at high clock frequencies to maximize the length of device idle intervals in order to apply DPM, but minimizing CPU energy involves lowering CPU clock frequencies.

This dissertation first illustrates how the DVS and DPM techniques can be integrated optimally for a real-time application potentially using multiple devices during execution. An exact characterization of the system-level energy as a function of the CPU frequency is provided. Using this characterization, an efficient static algorithm is designed to determine the CPU frequency and device transitioning decisions that minimize system-wide energy without violating the timing constraints.

Second, the integration of DVS and DPM for real-time applications that consist of multiple periodic tasks is considered. The problem of optimally using DPM for periodic real-time tasks, even in the absence of DVS, is formally shown to be NP-Hard in the strong sense. Then, a novel DPM framework called device forbidden regions is proposed and feasibility tests for both fixed- and dynamic-priority periodic real-time systems are developed. Using this framework as a building block, unified energy management frameworks that efficiently combine DVS and DPM at the system level are proposed.

Third, the problem of managing system-wide energy for periodic real-time tasks running on emerging chip-multiprocessor systems with global voltage and frequency constraint is addressed. Contributions made in this area include selecting the number of cores to execute the workload, and managing the global frequency at run-time across all cores to reduce dynamic energy while meeting the task deadlines.

A final contribution of this dissertation is the competitive analysis of online real-time scheduling problems under a given hard energy constraint. Specifically, worst-case performance bounds that apply to any online algorithm are derived, when compared to an optimal algorithm that has the knowledge of the input sequence in advance. Focusing on uniform value-density preemptive execution settings, optimal online and semi on-line algorithms achieving the best competitive factors are proposed. A number of additional fundamental results are provided for non-uniform value density, non-preemptive, and DVS-enabled execution settings.

Oral Defense of Doctoral Dissertation: Hardware-Assisted Protection and Isolation

Monday, July 18, 2011
2:30-4:30PM
Research 1, Room 401
Jiang Wang

Abstract

Software is prone to have bugs and vulnerabilities. To protect it, researchers normally go to a lower layer, such as protecting the applications from the kernel, or protecting the operating systems from the hypervisor, because the upper layer is controlled and depends on the lower layer. However, even a small hypervisor, which partitions the system hardware resources into different domains to support and isolate multiple virtual machines, may have some vulnerabilities and is hard to protect within itself. In this dissertation, we use hardware-assisted methods to monitor the integrity of the software running on top it. We present HyperCheck, a hardware-assisted tampering detection framework designed to protect the integrity of hypervisors or operating systems (OS). HyperCheck leverages the CPU System Management Mode (SMM), present in x86 systems and a dedicated commercial network card, to securely generate and transmit the full state of the protected machine to an external server. Using HyperCheck, we were able to ferret-out rootkits that targeted the integrity of both the Xen hypervisor and traditional OSes. Our experimental results show that HyperCheck can produce and communicate a scan of the state of the protected software in less than 40ms. Moreover, we analyze the attacks to the HyperCheck and similar SMM based protection systems and provide the defense mechanisms to those attacks. Besides the detection of the intrusion, another promising approach to protect the end user's computer is to separate the sensitive tasks such as financial related activities from the unsensitive tasks. For that purpose, we designed a system which has two operating systems installed: one is trusted and the other one is untrusted. The trusted OS only runs the trusted applications and is guaranteed to be separated from the untrusted OS. Without using a hypervisor, we leverage the commercial hardware and BIOS to enforce the isolation between the two OSes. Also, by utilizing the standard ACPI S3 sleep, we achieve a short delay when switching between the two OSes.

Oral Defense of Doctoral Dissertation: An Approach to Building Domain Specific Software Architectures Using Software Architectural Design Patterns

Thursday, July 21, 2011
10:30 AM
Eng 2302
Julie S. Fant

Abstract

Software architectural design patterns represent best practice solutions to common design challenges. However, applying design patterns in practice can be difficult because they are typically documented to be domain independent. This makes applying them in a particular domain difficult. Knowing where and at what level of abstraction software architectural design patterns should be applied in a given domain is not always clear. Currently, there are no existing approaches for building and validating domain specific software architectures that focus on reusing and composing existing software architectural design patterns. This dissertation addresses this gap by developing a software product line (SPL) based approach to building and validating domain specific software architectures from software architectural design patterns.

The key contributions of this research include: the definition of distributed real-time and embedded (DRE) executable design patterns; the definition of a SPL design approach that captures SPL variability at a higher degree of granularity using design patterns; the definition of different levels of required executable design pattern customizations; and a feature and design pattern based functional validation approach. Additionally, a domain specific SPL and two real world case studies are provided to validate and demonstrate the applicability of this approach.

Oral Defense of Doctoral Dissertation: A Congestion Pricing Model to Handle "Day of Operations" Airport Capacity Reductions due to Inclement Weather

Thursday, July 28, 2011
10:00AM
Eng 4801
Abdul Qadar Kara

Abstract

The Airline Industry in the United States is one of the major transportation alternatives for the US. It is a highly connected network used by over 712.6 million people in 2010 and is the most common means of travel for origin-destination pairs of greater than 250 miles. This has forced the US Department of Transportation (DOT) to consider solutions to relieve this congestion and to provide alternate solutions to the current system that has not been able to resolve this issue.

Recently researchers have been investigating market-clearing mechanisms such as congestion pricing to relieve short term congestion effectively. In such a price-based system, prices are announced for the scarce resources (e.g., runway access) and the users respond by either paying the announced price or opting out and either delaying or canceling the flight. Thus, by charging airlines to use the scarce runway access, the aim is to provide service to those who value it the most. For Ground Transportation, congestion pricing currently has been successfully implemented in several sectors.

This research proposes a system that implements the basic theory of congestion pricing and uses actual recorded operating costs of airlines to determine which airlines will (i) likely pay the price announced and fly, (ii) choose to delay the flight and eventually fly the flight in a less-congested time period, or (iii) cancel the flight. The research provides a new mechanism for calculating airline costs of delay as well as a mechanism for setting the congestion prices.

The results of imposing congestion prices are compared to other suggested rationing schemes to see the impact on airline costs, passenger throughput and passenger delay. The two alternative schemes chosen are Ration-by-Schedule (currently used by Air Traffic Management) and Ration-by-Distance (an approach that better reflects the airline's wish to fly their longest, most profitable flights).

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

Foundation Test-Out Exams: INFS 501 & INFS 515

Monday, August 22, 2011
2:00pm-4:30pm
ENGR 4201
INFS 501 2:00pm-3:00pm
INFS 515 3:30pm-4:30pm
REGISTRATION IS REQUIRED

Abstract

Please email your intent to csadmin@cs.gmu.edu. Include your name, G number and the exam(s) you wish to register for. A photo ID must be presented on the day of the exam. Each exam will be one hour in length.

It is important to note that you will be permitted to take this exam one time only. Failure to pass the exam will mean that you MUST take the foundation classes before enrolling in any core curriculum course.

CS PhD Program Orientation

Thursday, August 25, 2011
11:00AM - 12:00PM
Eng 4201
Prof. Ami Motro, Director (CS PhD program)

GTA Orientation

Thursday, August 25, 2011
12:00-2:00 pm
Eng 4201
Dr. Pearl Wang, Associate Chair

Rescheduled Foundation Test-Out Exam: INFS 519 & SWE 510

Thursday, August 25, 2011
2:00pm-4:30pm
ENGR 4801
INFS 519 2:00pm-3:00pm
SWE 510 3:30pm-4:30pm
REGISTRATION IS REQUIRED
(N/A if you registered for Tuesday 8/23/11)

Abstract

Please email your intent to csadmin@cs.gmu.edu. Include your name, G number and the exam(s) you wish to register for. A photo ID must be presented on the day of the exam. Each exam will be one hour in length.

It is important to note that you will be permitted to take this exam one time only. Failure to pass the exam will mean that you MUST take the foundation classes before enrolling in any core curriculum course.

SANG Seminar: Inter-Datacenter Bulk Transfers with NetStitcher

Thursday, August 25, 2011
11:00 am
Research 1, Room 401
Michael Sirivianos PhD CS, Duke University

Abstract

Large datacenter operators with sites at multiple locations dimension their key resources according to the peak demand of the geographic area that each site covers. The demand of specific areas follows strong diurnal patterns with high peak to valley ratios that result in poor average utilization across a day. In this talk I will show how to rescue unutilized bandwidth across multiple datacenters and backbone networks and use it for non-real-time applications, such as backups, propagation of bulky updates, and migration of data. Achieving the above is non-trivial since leftover bandwidth appears at different times, for different durations, and at different places in the world.

To this end, we have designed, implemented, and validated NetStitcher, a system that employs a network of storage nodes to stitch together unutilized bandwidth, whenever and wherever it exists. It gathers information about leftover resources, uses a store-and-forward algorithm to schedule data transfers, and adapts to resource fluctuations.

We have compared NetStitcher with other bulk transfer mechanisms using both a testbed and a live deployment on a real CDN. Our testbed evaluation shows that NetStitcher outperforms all other mechanisms and can rescue up to five times additional datacenter bandwidth thus making it a valuable tool for datacenter providers. Our live CDN deployment demonstrates that our solution can perform large data transfers at a substantially lower cost than naive end-to-end or store-and-forward schemes.

Speaker's Bio

Michael Sirivianos is a Junior Researcher at Telefonica Research in Barcelona. He earned a PhD degree in Computer Science from Duke University in 2010. I received a Diploma in Electrical and Computer Engineering from the National Technical University of Athens in 2002, and an M.S. degree in Computer Science from the University of California, San Diego in 2004. His research interests include cooperative content distribution, human verifiable secure device pairing, and introducing social trust in distributed system design.

GTA Teaching Workshop

Tuesday, August 30, 2011
10:30am-2:00pm
Eng 4201
Dr. Bethany Usher & Dr. Joshua Eyler, Associate Directors, CTE; Dr. Mary Zamon, Associate Director, OIA

CS Undergraduate Students Welcome BBQ!

Thursday, September 15, 2011
12:00-2:00 pm
Research Hall, Rm. 163
Ticket pick-up ENGR 4300

SWE Seminar: Test Case Selection Strategies for Model-Based Testing: Search-based Approaches and Industrial Case Study

Thursday, September 29, 2011
12:00-1:00 pm
Eng 4201
Lionel C. Briand

Abstract

Systems in all industry sectors increasingly rely on software for critical and complex functions. Software dependability must be ensured through verification and one of the most widespread and practical verification techniques is testing, that is the systematic and controlled execution of the system being verified. In recent years, Model-Based Testing (MBT) has attracted an increasingly wide interest from industry and academia. MBT allows automatic generation of a large and comprehensive set of test cases from system models (e.g., state machines), which leads to systematic system testing. However, even when using simple test strategies, applying MBT in large industrial systems often leads to generating large sets of test cases that cannot possibly be executed within time and cost constraints. In this situation, test case selection techniques must be employed to select a subset from the entire test suite such that the selected subset conforms to available resources while maximizing fault detection. In this talk, I will present the results of a comprehensive investigation involving alternative selection strategies, that are based on various heuristics and algorithms, and that attempt to maximize diversity or coverage in test suites. Based on an industrial case study, we will also estimate the potential benefits that can result from such test case selection strategies.

Speaker's Bio

Lionel C. Briand is heading software verification and validation activities at Simula Research Laboratory, Norway, where he is leading the newly established Certus research center and projects in collaboration with industrial partners. He is also a professor at the University of Oslo (Norway). Before that, he was on the faculty of the department of Systems and Computer Engineering, Carleton University, Ottawa, Canada, where he was full professor and held the Canada Research Chair (Tier I) in Software Quality Engineering. He has also been the software quality engineering department head at the Fraunhofer Institute for Experimental Software Engineering, Germany, and worked as a research scientist for the Software Engineering Laboratory, a consortium of the NASA Goddard Space Flight Center, CSC, and the University of Maryland, USA. Lionel has been on the program, steering, or organization committees of many international, IEEE and ACM conferences.

He is the coeditor-in-chief of Empirical Software Engineering (Springer) and is a member of the editorial boards of Systems and Software Modeling (Springer) and Software Testing, Verification, and Reliability (Wiley). He was on the board of IEEE Transactions on Software Engineering from 2000 to 2004. Lionel was elevated to the grade of IEEE Fellow for his work on the testing of object-oriented systems. His research interests include: model-driven development, testing and verification, search-based software engineering, and empirical software engineering.

SANG Seminar: Energy-Aware Standby-Sparing Technique for Periodic Real-Time Applications

Friday, September 30, 2011
12:00-1:00 pm
Eng 4201
Mohammad Haque, GMU PhD Student

Abstract

In this talk, we present an energy-aware standby-sparing technique for periodic real-time applications. A standby-sparing system consists of a primary processor where the application tasks are executed using Dynamic Voltage Scaling (DVS) to save energy, and a spare processor where the backup tasks are executed at maximum voltage/frequency, should there be a need. In our framework, we employ Earliest-Deadline-First (EDF) and Earliest-Deadline-Late (EDL) scheduling policies on the primary and spare CPUs, respectively. The use of EDL on the spare CPU allows delaying the backup tasks on the spare CPU as much as possible, enabling energy savings. We develop static and dynamic algorithms based on these principles, and evaluate their performance experimentally. Our simulation results show significant energy savings compared to existing reliability-aware power management (RAPM) techniques for most execution scenarios.

Speaker's Bio

Mohammad Atiqul Haque is a PhD student at George Mason University. His research interest includes real-time systems and low power computing. He is focusing on problems of fault-tolerant and energy-aware scheduling of real-time systems. His research advisor is Prof. Hakan Aydin. He completed his BSc in CSE from Bangladesh University of Engineering and Technology in 2006. He received his MS in CS from George Mason University in 2010.

Grand Seminar: The Role of Cheminformatics in Modern Drug Discovery

Tuesday, October 04, 2011
12:00 PM
Eng 4201
Simon Wang, Assistant Professor, Howard University

Abstract

Cheminformatics (also known as chemoinformatics and chemical informatics) is the application of informational techniques to a range of problems in the field of chemistry. These in silico techniques have played an increasing role in modern drug discovery and translational sciences in recent years. The development of cheminformatics methods and procedures that enable the automatic identification and extraction of privileged structures is very important in the context of generating knowledge from High-Throughput Screening (HTS) data. In this talk, I am going to introduce our recent efforts on the methodology development in this area that aims to improve the performance of virtual screening, i.e., finding molecular structures that are similar in their activity to the probe molecules or even predicting the activities of compounds in a library. Several successful cases using the cheminformatics technique will be presented as well.

Speaker's Bio

Simon Wang is currently an Assistant Professor and the Head of the Laboratory of Cheminfomatics and Drug Design at the Department of Pharmaceutical Sciences, School of Pharmacy Howard University (HU). He is also a faculty member for the Georgetown-Howard Universities Center for Clinical and Translational Science (GHUCTSA) Biomedical Informatics (BI) component, and an investigator for the District of Columbia Developmental Center for AIDS Research (DC D-CFAR). Dr. Wang received his B.S. degree in Pharmacy from Peking University School of Pharmaceutical Sciences, a M.S. degree in Pharmacology from Peking Union Medical College, and a Ph.D. degree in Computational Chemistry from the Department of Chemistry and Quantum Theory Project at the University of Florida. Prior to his joining the HU in late 2010, Dr. Wang had postdoctoral training with Dr. Harel Weinstein at Cornell University and had been a junior faculty at the Eshelman School of Pharmacy, University of North Carolina at Chapel Hill (UNC-CH).

Grand Seminar: Mobile Robot Perception for Long-term Autonomy

Wednesday, October 05, 2011
2:00 PM
Eng 4201
Gabe Sibley, Assistant Professer, George Washington University

Abstract

If mobile robots are to become ubiquitous, we must first solve fundamental problems in perception. Before a mobile robot system can act intelligently, it must be given -- or acquire -- a representation of the environment that is useful for planning and control. Perception comes before action, and the perception problem is one of the most difficult we face.

An important goal in mobile robotics is the development of perception algorithms that allow for persistent, long-term autonomous operation in unknown situations (over weeks or more). In our effort to achieve long-term autonomy, we have had to solve problems of both metric and semantic estimation. In this talk I will describe two recent and interrelated advances in robot perception aimed at enabling long-term autonomy.

The first is relative bundle adjustment (RBA). By using a purely relative formulation, RBA addresses the issue of scalability in estimating consistent world maps from vision sensors. In stark contrast to traditional SLAM, I will show that estimation in the relative framework is constant-time, and crucially, remains so even during loop-closure events. This is important because temporal and spatial scalability are obvious prerequisites for long-term autonomy.

Building on RBA, I will then describe co-visibility based place recognition (CoVis). CoVis is a topo-metric representation of the world based on the RBA landmark co-visibility graph. I will show how this representation simplifies data association and improves the performance of appearance based place recognition. I will introduce the "dynamic bag-of-words" model, which is a novel form of query expansion based on finding cliques in the co-visibility graph. The proposed approach avoids the -- often arbitrary -- discretization of space from the robot's trajectory that is common to most image-based loop-closure algorithms. Instead, I will show that reasoning on sets of co-visible landmarks leads to a simple model that out-performs pose-based or view-based approaches, in terms of precision and recall.

In summary, RBA and CoVis are effective representations and associated algorithms for metric and semantic perception, designed to meet the scalability requirements of long-term autonomous navigation.

Speaker's Bio

Gabe Sibley is a robotics scientist and assistant professor in Computer Science at George Washington University. He was formerly in the University of Oxford in the Mobile Robotics Group. He did his PhD at the University of Southern California and at NASA-JPL, where he worked on long-range data-fusion algorithms for planetary landing vehicles, unmanned sea vehicles and unmanned ground vehicles. His core interest is in probabilistic perception algorithms and estimation theory that enable long-term autonomous operation of mobile robotic systems, particularly in unknown environments. He has extensive experience with vision based, real-time localization and mapping systems, and is interested in fundamental understanding of sufficient statistics that can be used to represent the state of the world. His research uses real-time, embodied robot systems equipped with a variety of sensors -- including lasers, cameras, inertial sensors, etc. -- to advance and validate algorithms and knowledge representations that are useful for enabling long-term autonomous operation.

Computer Game Design Lecture Series: The Whoosh Moment

Wednesday, October 05, 2011
7:00 PM
Research I, Rm. 163
Jerry Tessendorf, Prof. Clemson University

Abstract

Complex and massive computations are performed every day in the effort to make movies that audiences want to see. The quality of computer graphics are so good that many times they cannot be detected even by the expert practitioners. Other times the visual effects are obvious to anyone. Yet the focus of this technology is to achieve an artistic vision in storytelling. In this talk we look at the details of computer graphics and art that the experts create every day, using some examples from big (and not so big) feature films. We see what was done practically on set, what was done with the magic of computer graphics, and how people come to learn these techniques and join the small group of practitioners. When finished, you will never again be able to watch movies the way you did before.

Speaker's Bio

Jerry Tessendorf is a Professor of Visual Computing, and Director of the Digital Production Arts program at Clemson University. His research is in fluid dynamics, radiative transfer, and production workflow for feature films. He has developed new movie production techniques and software for 15 years at Rhythm & Hues and Cinesite Digital Studios, and received an Academy Award for Technical Achievement. He has a Ph.D. in physics from Brown University.

SANG Seminar: Maximal-utility Rate Allocation for Energy Harvesting Wireless Sensor Networks

Friday, October 07, 2011
12:00-1:00 PM
Eng 4201
Bo Zhang, PhD Student, CS

Abstract

There is currently tremendous interest in deploying energy harvesting wireless sensor networks. Engineering such systems requires striking a careful balance between sensing performance and energy management. Our work addresses this problem through the design and analysis of a harvesting aware utility-based sensing rate allocation algorithm. Based on a network utility formulation, we show that our algorithm is optimal in terms of assigning rates to individual nodes to maximize overall utility, while ensuring energy-neutral operation. To our knowledge, our work is the first optimal solution that maximizes network utility through rate assignments for tree-structured energy harvesting sensor networks. Our algorithm is fast and efficient with running time O(N3), where N is the number of nodes. We evaluate the performance, scalability, and overhead of our algorithm for various utility functions and network sizes, underlining its significant advantages.

Speaker's Bio

Bo is a PhD student in computer science department of George Mason University. He is currently working on energy and performance management for energy harvesting wireless sensor networks. He earned his BS. degree from Huazhong University of Science and Technology, China; MS. degree from University of Cincinnati.

Volgenau School Seminar: Smart & Mobile Devices in Foreign Wars: Locking Down Linux, Software Apps, and Communications

Friday, October 07, 2011
11:00 AM
Research I, Rm. 163
Angelos Stavrou, CS Department

Abstract

Recent advances in the hardware capabilities of mobile hand-held devices have fostered the development of open source operating systems and a wealth of applications for mobile phones and table devices. This new generation of smart devices, including iPhone and Google Android, are powerful enough to accomplish most of the user tasks previously requiring a personal computer. In this talk Dr. Stavrou will discuss the cyber threats that stem from these new smart device capabilities and the online application markets for mobile devices. These threats include malware, data exfiltration, exploitation through USB, and user and data tracking. He will present the ongoing GMU and NIST efforts to defend against or mitigate the impact of attacks against mobile devices. Our approaches involve analyzing the source code and binaries of mobile applications, hardening the Android Kernel, using Kernel- level network and data encryption, and controlling the communication mechanisms for synchronizing the user contents with computers and other phones. Dr. Stavrou will also explain the enhanced difficulties in dealing with these security issues when the end-goal is to deploy security-enhanced smart phones into military combat settings. The talk will conclude with a discussion of his current and future research directions and outcomes in. Dr. Angelos Stavrou is an Assistant Professor in the Computer Science Department and a member of the Center for Secure Information Systems at George Mason University, Fairfax, Virginia. His current research interests include security and reliability for distributed systems, security principles for virtualization, and anonymity with a focus on building and deploying large-scale systems.

CS Dept Colloquium: Searching in the "Real World"

Wednesday, October 12, 2011
11:00 AM
Eng 4201
Ophir Frieder, Georgetown University

Abstract

For many, "searching" is considered a mostly solved problem. In fact, for text processing, this belief is factually based. The problem is that most "real world" search applications involve "complex documents", and such applications are far from solved. Complex documents, or less formally, "real world documents", comprise of a mixture of images, text, signatures, tables, etc, and are often available only in scanned hardcopy formats. Search systems for such document collections are currently unavailable.

We describe our efforts at building a complex document information processing prototype. This prototype integrates "point solution" (mature) technologies, such as document readability enhancement, OCR capability, signature matching and handwritten word spotting techniques, search and mining approaches, among others, to yield a system capable of searching "real world documents". The described prototype demonstrates the adage that "the whole is greater than the sum of its parts". Our complex document benchmark development efforts are likewise presented.

Having described the global approach, we describe some point solutions which we developed over the years. These include an image enhancer, an Arabic stemmer, and a natural language source integration fabric called the Intranet Mediator.

Speaker's Bio

Ophir Frieder holds the Robert L. McDevitt, K.S.G., K.C.H.S. and Catherine H. McDevitt L.C.H.S. Chair in Computer Science and Information Processing and is Chair of the Department of Computer Science at Georgetown University. He is a Fellow of the AAAS, ACM, and IEEE.

SANG Seminar: Mobile MapReduce: Minimizing Response Time of Computing Intensive Mobile Applications

Friday, October 14, 2011
12:00-1:00 PM
Eng 4201
Mohammad Hassan, GMU CS PhD Student

Abstract

The increasing popularity of mobile devices calls for effective execution of mobile applications. A lot of research has been conducted on properly splitting and outsourcing computing intensive tasks to exter- nal resources (e.g., public clouds) by considering insufficient computing resources on mobile devices. However, little attention has been paid to the overall users' response time, where the network may dominate. In this talk, we will discuss how we can effectively minimize users' response time for mobile applications. We consider both the impact of the network and the computing itself. We first show that outsourcing to nearby residential computers may be more advantageous than pub- lic clouds for mobile applications due to network impact. Furthermore, to speed up computing, we leverage parallel processing techniques. Ac- cordingly, we propose to build Mobile MapReduce (MMR) to effectively execute outsource computing intensive mobile applications. Based on the original MapReduce framework, a new scheduling model is built in MMR that can always leverage the best computing resources to conduct computation with appropriate parallel processing. To demonstrate the performance of MMR, we run several real-world applications, such as text searching, face detection, and image processing, on the prototype. The results show great potentials of MMR in minimizing the response time of the outsourced mobile applications.

Speaker's Bio

Mohammed A. Hassan is a PhD student at George Mason University. His research interest includes mobile-cloud computing. He completed his BSc in CSE from Bangladesh University of Engineering and Technology in 2006.

Volgenau School Seminar: Mining Massive Time Series Databases

Monday, October 17, 2011
2:00 PM
Johnson Center, Ground Level, Gold Room
Jessica Lin, PhD, CS Department

Abstract

Much of the world's supply of data is in the form of time series. One obvious problem of handling time series databases concerns with its typically massive size—gigabytes or even terabytes are common, with more and more databases reaching the petabyte scale. Most classic data mining algorithms do not perform or scale well on time series data due to their unique structure. In particular, the high dimensionality, very high feature correlation, and the typically large amount of noise that characterize time series data present a difficult challenge. As a result, time series data mining has attracted an enormous amount of attention in the past two decades. This presentation gives an overview of my contributions in the field of time series data mining. The emphasis of my research is on the discovery of important patterns in time series data. These significant patterns can manifest either as frequently encountered (or repeated) patterns, rare (or anomalous) patterns, or latent structure. The previous body of work in this area has been mostly concentrated on the identification of previously known patterns. The major distinction of my work is that it offers the ability to discover important, unknown patterns in an effective and efficient manner. I will also discuss the broad impact of my work in various domains, including medicine, manufacturing, astronomy, defense, and earth sciences. Jessica Lin is an Assistant Professor in the Department of Computer Science at George Mason University. She received her PhD degree from University of California, Riverside in June, 2005. Her research interests are in data mining, databases, information retrieval, and machine learning. Her research is partially funded by US Army and Intel Corporation.

CS Dept Colloquium: New Perception of "Computer Science"

Monday, October 17, 2011
11:00 AM
Eng 4201
Jozef Gruska, Faculty of Informatics, Masaryk University, Brno, Czech Republic

Abstract

The talk will present a new, much broader and deeper, vision of the field usually called "Computer Science" or "Informatics" and will justify why such a new view is possible and much needed.

New view is to a large extent motivated by the recent developments in "natural computing" (quantum and biological) and the fact that many areas of science and technology are information-processing driven. Understanding this broader and deeper view of Informatics puts the field into the position of the main servant and also guiding and inspiring force for other areas of science and technology with broad impacts almost everywhere. This new perception of the Informatics sees the field as having four closely related components: scientific, engineering, methodological and applications, which are described and illustrated by their grand challenges.

As an area of science, Informatics is having similar goals as other areas of basic science and represents a new window to see and explore the real and artificial worlds and lives. A special attention is given to the new and very powerful methodology Informatics provides that much extends the role mathematics used to have in serving sciences, technologies and other areas of human activities.

Speaker's Bio

Prof. Jozef Gruska got his PhD from Slovak Avademy of Sciences in Bratislava (1966) Slovakia, in Computer Science. Currently he is a professor at the Faculty of Informatics with Masaryk University in Brno, Czech Republic. He is an expert in Theoretical Informatics and an author of two foundation monographs on "Foundations of Computing" US, 1997 and "Quantum Computing", UK, 1999 and has over 150 publications. He is an elected member of the Academia Europaea, member of its Informatics Section committee (since 2006) and member of its Council (2007-2010). "Computer pioneer" award of IEEE USA and many other awards and recognitions including Bolzano award of the Czech Academy of Sciences. Currently he is the head of two big interdisciplinary projects at Masaryk University.

He is a founder of seven regular international conferences including Asia Quantum Information Science (Japan, Korea, China, India), 2001-2013. He gave more than 250 talks on international events including two invited talks at the World Computer Congress. He was the founding head (1989-1996) of the main international committee for theoretical informatics at IFIP (International Federation of Information Processing). He spent more than 15 years at major universities of Europe and North America.

Volgenau School Seminar: Challenges in Mixed-Reality Systems

Tuesday, October 18, 2011
10:00 AM
Johnson Center, 3rd Floor, Room A
Joao Sousa, PhD, CS Department

Abstract

Mixed-reality systems merge the virtual and the real. The real world may be presented to users augmented with views of virtual and real entities albeit distant or hidden from direct sight. The virtual helps monitor and control the real world, via sensors and actuators, with applications in training and entertainment, control (home automation, surveillance, energy management, transportation, assisted living...) search and rescue, public safety, defense, disaster response, etc. This talk discusses a number of challenges that I and my students have been working on, namely concerning software architectures and middleware, end-user programming, decentralized context-awareness, security and privacy, and self- adaptation. João joined the Department of Computer Science at GMU as an assistant professor in 2006. He obtained his PhD in CS (2005) and MS in Software Engineering (2000) from Carnegie Mellon University, after working for 10 years as a software architect and project manager for the financial industry.

Volgenau School Seminar: Expressive Probablistic Logic for Knowledge Fusion

Thursday, October 20, 2011
11:00 AM
Johnson Center, 3rd Floor, George's
Kathryn Blackmond Laskey, PhD, SEOR Department

Abstract

In today's interconnected world, we are reminded constantly of the need to "connect the dots," combining information from disparate sources to identify and head off potential threats. Advances in the ability to sense, communicate, store and process data have given rise to a deluge of raw data that must be fused and transformed into actionable knowledge. Although theory and algorithms for low-level data fusion have matured rapidly, current approaches to knowledge-level fusion still depend heavily on manual processing. Techniques for making semantic information explicit and computationally accessible are key enablers for knowledge-level fusion. Because of the pervasiveness of uncertainty, these methods must properly account for and manage uncertainty. This presentation describes Multi-entity Bayesian networks (MEBN), a computational logic designed to address the challenges of knowledge-level fusion. MEBN combines the expressive power of first-order logic with the ability of Bayesian networks to represent and reason with uncertainty. A MEBN domain model implicitly represents a joint probability distribution over situations involving unbounded numbers of interrelated and interacting entities. The application of MEBN to problems in knowledge-level fusion is discussed. Kathryn Blackmond Laskey is Associate Professor of Systems Engineering and Operations Research (SEOR) and Associate Director of the Center of Excellence in Command, Control, Communications, Computing and Intelligence (C4I Center). Her research focuses on extending traditional knowledge representation methods to represent uncertainty, and applying Bayesian methods to automated support for multi-source fusion and situational awareness. She received her PhD degree in statistics and public policy from Carnegie Mellon University, MS in mathematics from the University of Michigan, and BS in mathematics from the University of Pittsburgh.

SWE Seminar: Variability Modeling and Meta-Modeling for Model Driven Service Oriented Architectures

Monday, October 24, 2011
12:00 PM
Eng 4201
Mohammed A. Abu-Matar, Ph.D. Candidate

Abstract

Service Oriented Architecture (SOA) has emerged as an architectural style for distributed computing that promotes flexible deployment and reuse. One of the major benefits claimed for SOA is the flexible building of IT solutions that can react to changing business requirements quickly and economically. Services could be consumed by many applications that have different requirements. In addition, applications usually change by adding new requirements, removing existing requirements, or updating existing requirements. Thus, applications that consume the same service usually exhibit varying requirements. Varying requirements usually necessitate varying software architectures that satisfy the varying requirements of software applications. Thus, both requirements and architectures have intrinsic variability characteristics. SOA development practices currently lack a systematic approach for managing variability in service requirements and architectures. This talk reports on research that addresses this gap by introducing a framework for managing variability in SOA in a systematic and unified way. The research introduces an approach to model SOA variability with a multiple-view service variability model and a corresponding meta-model. The research integrates Software Product Lines (SPL) concepts with the different service views using UML and SoaML. The research argues that the multiple-view service variability modeling approach facilitates variability modeling of service application families in a systematic and platform independent way. The key contributions of this research include: Multiple-View Service Variability Meta-Model, Multiple-View Service Variability Model, Consistency Checking and Mapping Rules, Model Driven Framework for Service-Oriented SPLs, Service Member Applications Derivation Rules, Explicit Modeling of Service Coordination Variability, and a prototype that realizes the introduced framework.

Speaker's Bio

Mohammad Abu Matar is a software engineering academic and practitioner with over 17 years of technical experience in teaching, research, management, architecture, systems engineering, training, software design and development. Mohammad has earned a BS in Electrical Engineering (Wright State University), an MS in Information Technology (Regis University), an MS in Software Engineering (George Mason University), and he is a PhD candidate in Software Engineering at George Mason University. Mohammad’s specialty is the architecture of multi-tier distributed software systems with a special interest in Service Oriented Architecture (SOA). Mohammad is an Affiliate Adjunct Faculty at the MS in Software Engineering program of Regis University (Denver, Colorado). In addition, Mohammad has developed and delivered many technical corporate training sessions in software design, architecture, and SOA.

SANG Seminar: Optimizing Energy Consumption under Flow and Stretch Constraints

Friday, October 28, 2011
12:30-1:30 PM
Eng 4201
Zhi Zhang, Ph.D. Candidate GMU

Abstract

In embedded systems and data-center systems (systems, for short), it is widely accepted that energy consumption has become the bottleneck of system's performance improvement and it is one of the most significant factors to optimize. Unfortunately, an effective energy-aware strategy usually has an adverse impact on a job's flow time or stretch --- two important user-perspective system performance metrics. In some cases, the more energy is saved by an energy-aware policy, the more flow time and the larger stretch occur to jobs. In this paper, we investigate the impact on job processing delay introduced by power-down energy-saving mechanisms. Specifically, we study bicriteria algorithms that minimize maximum flow time or largest stretch under a fixed energy budget and minimize total energy consumption under an upper bound of flow time or stretch. We develop optimal offline algorithms to quantitatively balance the system-perspective performance metric (energy consumption) and the user-perspective performance metric (flow time and stretch). We also develop two simple min-energy online algorithms against weakened adversaries. We prove that (1.) with appropriate extra flow time, an online algorithm can beat a non-idling offline algorithm, which achieves the minimum flow time, in terms of the total energy required; (2.) an optimal deterministic online algorithm, in terms of competitive ratio with respect to energy consumption, has a bounded times of optimal stretch.

Speaker's Bio

Zhi Zhang is a PhD candidate at George Mason University. Her research focuses on algorithm design and analysis, mainly on applied scheduling algorithms and green algorithms. Zhi received her BS and MS in Computer Science from Wuhan University, China in 2006 and 2008, respectively.

MS Thesis Defense: Local Minima Hopping Along the Protein Energy Surface

Thursday, November 03, 2011
11:00 AM
Eng 4201
Brian Olson, GMU MS CS Student

Abstract

Modeling of protein molecules in silico for the purpose of elucidating the three-dimensional structure where the protein is biologically active employs the knowledge that the protein conformational space has an underlying funnel-like energy surface. The biologically-active structure, also referred to as the native structure, resides at the basin or global minimum of the energy surface. A common approach among computational methods that seek the protein native structure is to search for local minima in the energy surface, with the hope that one of the local minima corresponds to the global minimum. Typical stochastic search methods, however, fail to explicitly sample local minima. This thesis proposes a novel algorithm to directly sample local minima at a coarse-grained level of detail. The Protein Local Optima Walk (PLOW) algorithm combines a memetic approach from evolutionary computation with cutting-edge structure prediction protocols in computational biophysics. PLOW explores the space of local minima by explicitly projecting each move at the global level to a nearby local minimum. This allows PLOW to jump over local energy barriers and more effectively sample near-native conformations. An additional contribution of this thesis is that the memetic approach in PLOW is applied to FeLTr, a tree-based search framework which ensures geometric diversity of computed conformations through projections of the conformational space. Analysis across a broad range of proteins shows that PLOW and memetic FeLTr outperform the original FeLTr framework and compare favorably against state-of-the-art ab-initio structure prediction algorithms.

SANG Seminar: An Empirical Evaluation of Battery Power Consumption for Streaming Data Transmission to Mobile Devices

Friday, November 04, 2011
12:30-1:30 PM
Eng 4801
Yao Liu, GMU CS PhD Student

Abstract

Internet streaming applications are becoming increasingly popular on mobile devices. However, receiving streaming services on mobile devices is often constrained by their limited battery power supply. Various techniques have been proposed to save battery power consumption on mobile devices, mainly focusing on how much data to transmit and how to transmit.

In this study, we conduct an experiment-based study with 11 Internet streaming applications using different streaming protocols. Our goal is to empirically investigate the battery power consumption on the wireless network interface for receiving streaming data via different approaches. Through measurement and analysis, we find that (1) the Chunk-based streaming is widely used in practice and it is most power-efficient because the traffic shaping technique is adopted to utilize PSM on mobile devices to save battery power consumption; however, it may cause quality degradation from time to time; (2) reducing streaming data transmission (by switching to a lower streaming quality) can marginally help save battery power consumption in RTSP, Pseudo streaming, and Chunk-based streaming applications; but it is effective for P2P streaming applications; (3) P2P streaming to mobile devices is not power-efficient because of the additional transmission of control traffic and uploading traffic; and reducing upload alone does not help for battery power saving.

Speaker's Bio

Yao Liu is a Ph.D. student of Computer Science Department at George Mason University. Her research interests include Internet and mobile content delivery, multimedia systems and networking.

Oral Defense of Doctoral Dissertation: Trust Management in Smart Spaces

Friday, November 04, 2011
8:30 - 10:30 AM
Eng 4801
Dalal Ahmed Alarayed, PhD Candidate Computer Science

Abstract

Smart spaces introduce new issues that are not addressed sufficiently by available trust management models. With the ultimate goal of producing a generalized model that is useable in diverse problem domains including smart space scenarios, a trust model designed for smart spaces was devised. Trust in Smart Spaces (TISS) divides participants into categories to facilitate for user defined constants to be plugged into it to implement stereotyping, a mechanism by which decision makers trust trustees to varying degrees based on their category membership. In addition, TISS handles multi-link trust decisions, in which multiple trustees exist in a single trust decision, and multilateral decision making where more than one trustor participate in the decision making. Personalization, through the use of privacy policies, enables individual decision makers to tailor the use of the trust model to their preferences based various criteria such as the location of the trustee, the location of the trustor, the trustee’s category membership, conditions on time and conditions on other stored trust data. General Trust Management-GTM is a generalization of TISS that is applicable in multiple problem domains such as packet routing, service provider selection, content filtering, granting access and location disclosure. GTM combines functionality with versatility. It explicitly handles trust management functions including trust formation, dissemination and evolution. GTM preserves all TISS’s functionality including the ability to handle multi-link trust decisions. It applies stereotyping by dividing participants into user defined categories, and personalization through the use of privacy policies. GTM’s generality is validated by demonstrating its applicability in diverse problem domains. To investigate whether end users can understand and use TISS, An Android prototype of TISS was built to target a location disclosure scenario. A user study was conducted on a sample of users that used the prototype and gave feedback on TISS’s usability.

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

Oral Defense of Doctoral Dissertation: Variability Modeling and Meta-Modeling for Model Driven Service Oriented Architectures

Thursday, November 10, 2011
1:00-3:00 PM
Eng 4201
Mohammed A. Abu-Matar, Ph.D. Candidate

Abstract

Service Oriented Architecture (SOA) has emerged as an architectural style for distributed computing that promotes flexible application development and reuse. One of the major benefits claimed for SOA is the flexible building of IT solutions that can react to changing business requirements quickly and economically. Services could be consumed by many applications that have different requirements. In addition, applications usually change by adding new requirements, removing existing requirements, or updating existing requirements. Thus, applications that consume the same service usually exhibit varying requirements. Varying requirements usually necessitate varying software architectures that satisfy the varying requirements of software applications. Thus, both requirements and architectures have intrinsic variability characteristics.

SOA development practices currently lack a systematic approach for managing variability in service requirements and architectures. This dissertation addresses this gap by applying software product line (SPL) concepts to model SOA systems as service families. The dissertation introduces an approach to model SOA variability with a multiple-view service variability model and a corresponding meta-model. The approach integrates SPL concepts of feature modeling and commonality/variability analysis with multiple service requirements and architectural views by using UML and the Service Oriented Architecture Modeling Language (SoaML). At the heart of this research is a multiple-view meta-model that defines the relationships among variable service views and maps features to variable service models along with a corresponding consistency checking rules that ensure the consistency of the multiple service views as they change. The dissertation describes how to derive family member applications from the SPL and presents a validation of the approach. This dissertation makes the case that the presented multiple-view service variability modeling and meta-modeling approach facilitates variability modeling of service families in a systematic and platform independent way. The key contributions of this research include: Multiple-View Service Variability Meta-Model, Multiple-View Service Variability Model, Consistency Checking and Mapping Rules, Model Driven Framework for Service Oriented SPL Engineering, Service Member Application Derivation Rules, Explicit Modeling of Service Coordination Variability, and a proof-of-concept tool prototype.

Speaker's Bio

Mohammed A. Abu-Matar, Ph.D. Candidate, received a Bachelor of Science, Wright State University, 1993 and Master of Science, Regis University, 2003

SANG Seminar: Attack-resilient Data Aggregation in Wireless Sensor Networks

Friday, November 11, 2011
12:30-1:30 PM
Eng 4201
Sanjeev Setia, Chair CS Department

Abstract

In many sensor applications, the data collected from individual nodes is aggregated at a base station or host computer. To reduce energy consumption, many systems also perform in-network aggregation of sensor data at intermediate nodes enroute to the base station. Most existing aggregation algorithms and systems do not include any provisions for security, and consequently these systems are vulnerable to a wide variety of attacks. In this talk, I will discuss a particularly pernicious attack in which a compromised node falsifies the sub-aggregate transmitted to its ancestors in the aggregation hierarchy, and present efficient algorithms for computing the SUM and MEDIAN aggregates that are resilient to this attack.

Speaker's Bio

Sanjeev Setia is a Professor and Chair of the Computer Science Department at George Mason University. He received his PhD from the University of Maryland, College Park in 1993. His research interests are in ad hoc and sensor networks, network security and performance evaluation of computer systems. In recent years, he has worked extensively on security mechanisms and protocols for ad hoc and sensor networks. He was the founder of the ACM Workshop on Security in Ad hoc and Sensor Networks (SASN) and served as its co-organizer in 2003 and 2004. His research has been funded by NSF, NASA and DARPA.

SANG Seminar - Rescheduled: BlueStreaming: Towards Power-Efficient Internet P2P Streaming to Mobile Devices

Tuesday, November 22, 2011
12:30-1:30 PM
Eng 4201
Yao Liu, GMU CS PhD Student

Abstract

In addition to continuously receiving streaming data, a mobile device in popular Internet P2P streaming applications also needs to upload to other peers and exchange control information (e.g., buffermaps and file chunk requests) with neighbors for such download and upload. This leads to excessive battery power consumption on the mobile device.

In this study, we first conduct Internet experiments to study in-depth the impact of control traffic and uploading traffic on battery power consumption with several popular Internet P2P streaming applications. Motivated by measurement results, we design and implement a system called BlueStreaming that effectively utilizes the commonly existing Bluetooth interface on mobile devices. Instead of activating WiFi and Bluetooth interfaces alternatively, BlueStreaming keeps Bluetooth active all the time to transmit delay-sensitive control traffic while using WiFi for streaming data traffic. BlueStreaming trades Bluetooth's power consumption for much more significant energy saving from shaped WiFi traffic. To evaluate the performance of BlueStreaming, we have implemented prototypes on both Windows and Mac to access existing popular Internet P2P streaming services. The experimental results show that BlueStreaming can save up to 46 percent battery power compared to the commodity PSM scheme.

Speaker's Bio

Yao Liu is a Ph.D. student of Computer Science Department at George Mason University. Her research interests include Internet and mobile content delivery, multimedia systems and networking.

Oral Defense of Doctoral Dissertation: A Learning-Based Framework for Engineering Feature-Oriented Self-Adaptive Software Systems

Friday, December 02, 2011
1:00-3:00 PM
Eng 4801
Ahmed Elkhodary, CS PhD Candidate

Abstract

Self-adaptive software systems are capable of adjusting their behavior at runtime to achieve certain functional or quality of service goals. Often a representation that reflects the internal structure of the managed system is used to reason about its characteristics and make the appropriate adaptation decisions. However, in practice, self-adaptive software systems are often complex, dynamic, and the structure of managed system may not be completely known at design time. In addition, runtime conditions can radically change the internal structure in ways that were not accounted for during their design. As a result, unanticipated changes at runtime that violate the assumptions made about the internal structure of the system could make the adaptation decisions inefficient and inaccurate. In this dissertation, we present an approach for engineering self-adaptive software systems that brings about two innovations: (1) a feature-oriented approach for representing engineers’ knowledge of adaptation choices that are deemed practical, and (2) an online learning-based approach of assessing and reasoning about adaptation that does not require an explicit representation of the internal structure of the managed software system. Engineers’ knowledge, represented in feature-models, adds structure to online learning, which in turn makes online learning feasible. We present an empirical evaluation of a proof-of-concept implementation of the framework using a real world self-adaptive software system. Results demonstrate the framework’s ability to accurately learn the changing dynamics of the system, while achieving efficient analysis and adaptation.

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

Oral Defense of Doctoral Dissertation: Development of a Secure Mobile GPS Tracking and Management System

Friday, December 02, 2011
2:00 - 4:00 PM
Eng 2901
Anyi Liu, IT PhD Candidate

Abstract

With increasing demand of mobile devices and cloud computing, it is important to develop efficient mobile application and its secured backend, such as web applications and virtualization environment. This dissertation reports a systematic study of mobile application development and the security issues of its related backend. First, to standardize the software development of mobile application, an efficient mobile application that addresses the key issues of mobile application development has been designed. Second, to prevent against code-injection attacks towards web applications, a black-box input validation approach, which harnesses the effectiveness of genetic and input validation algorithms has been developed and implemented. Third, to protect user's private information from being exfiltrated to outside attacker through covert channels (CCs), an architectural solution has been developed to detect CCs in real-time. Our intrusion detection system can detect covert channels and does not require legitimate traffic to build normal behavior models. To detect more advanced covert channels, a novel metric has been developed to quantitatively measure the difference between the timing patterns of legitimate traffic and CCs. The evaluation demonstrates that our metric can quantitatively measure and detect covert channels that hidden in the outbound networking flow of VMs with high detection and a low false positive rate at runtime. A copy of this doctoral dissertation is on reserve at the Johnson Center Library.

SANG Seminar: SecureSwitch: BIOS-Assisted Isolation and Switch between Trusted and Untrusted Commodity OSes

Friday, December 02, 2011
12:30-1:30 PM
Eng 4201
Kun Sun, Center for Secure Information Systems

Abstract

Protecting commodity systems with commercial Operating Systems OSes) without significantly degrading performance or usability still remains an open problem. To make matters worse, the overall system security depends on desktop applications with complex code-bases that perform multiple and inter-dependent tasks often dictated by Internet-borne code. Recent research has shown the need for context-dependent trustworthy environments where the user can segregate different activities to lower risk of cross-contamination after an infection and safeguard private information.

We introduce a novel BIOS-assisted mechanism to enable secure instantiation and management of trusted execution environments, tailored to separate security-sensitive activities from untrusted ones. A key characteristic of our system is usability: the capability to quickly and securely switch between operating environments in a single physical machine without requiring any specialized hardware or OS and application code modifications. Our goal is to eliminate any mutable, non-BIOS code sharing while securely reusing existing hardware. We demonstrate that, even if the untrusted OS becomes compromised, there is no potential for exfiltration or inference attack against data in the trusted OS. To safeguard against OS spoofing attacks, we can force the user to physically set a hardware switch, an action that cannot be reproduced by software. In addition, we provide visible indication to the user about the current environment leveraging one of the front panel Light Emitting Diodes (LEDs). Using our prototype implementation, we measured the switching process to be approximately six seconds on average. This quick and user-friendly switching process empowers the user to frequently and seamlessly alternate between trusted and untrusted environments.

Speaker's Bio

Dr. Kun Sun is a Research Professor in the Center for Secure Information Systems (CSIS) at George Mason University. He received his Ph.D. in Computer Science from North Carolina State University in 2006. Before joining GMU, Dr. Sun was a Senior Research Scientist in Intelligent Automation Inc. From 2000 to 2001, Dr. Sun was a Member of the Technical Staff at Bell Labs, Lucent Technology. His current research focuses on trustworthy computing environment, moving target defense, enterprise-level security metrics, and security in MANET and wireless sensor networks.

Oral Defense of Doctoral Dissertation: Developing Enterprise Architectures to Address the Enterprise Dilemma of Deciding What Should Be Sustained Versus What Should Be Changed

Monday, December 05, 2011
10:00 AM - 12:00 PM
Eng 4201
J. Michael Harrell, GMU IT PhD Candidate

Abstract

Enterprise architecture is a relatively new concept that arose in the latter half of the twentieth century as a means of managing the information technology resources within the enterprise. The goal of this research has been to discover strategies that can attenuate the difficulties that result from wicked problems, complexity, and the enterprise learning curve and also improve the likelihood of developing an enterprise architecture that delivers a positive return on investment for the enterprise. Towards this goal, this research establishes: the focus and scope of enterprise architecture by defining the bounds of what enterprise architecture should address; develops a core set of enterprise business questions from which to begin enterprise architecture development; develops an enterprise architecture metamodel that supports enterprise architecture metamodel development; and develops a methodology that aids the enterprise architect in focusing the development effort on obtaining significant value while reducing the risk of expending resources developing architectural artifacts of little or no value.

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

Oral Defense of Doctoral Dissertation: Decision Support Framework for Legacy System Integration

Tuesday, December 06, 2011
2:00 - 4:00 PM
Eng, 3507
Robert J. Knapper, PhD IT Candidate

Abstract

Legacy software systems are often the subject of analyses in attempts to affect modifications to extend their useful lifetimes. For the systems engineer, reengineering legacy systems or integrating legacy systems with new development and/or off-the-shelf software are areas which may provide particular challenges and where effective software system analysis is critical. When analyzing a software system, a range of system artifacts may be used. These artifacts include a description of the system in natural language, program design language descriptions, flow charts, object diagrams, as well as the system code. All of these artifact types can contain equivalent information at their specific level of abstraction. However, deriving information that is needed for reengineering or integrating a legacy system from each of these artifacts cannot be performed equivalently. For example, analyzing a poorly structured system with little or no design documentation is a particularly challenging problem. Software modeling using an abstract architecture representation as a common description vehicle, such as the Unified Modeling Language or some other, can “level the playing field” by creating representations of disparate systems at the same level of abstraction. This is of importance to reengineering efforts; it shows promise for supporting and facilitating integration efforts between legacy and other systems. This dissertation presents a methodology to express and manipulate (i.e. restructure) system architecture-level models, an evaluation of the integratability of those models, and proposes a decision support system framework that makes the use of the methods.

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

Oral Defense of Doctoral Dissertation: Sharing Intelligently Derived Location Context While Preserving Privacy

Tuesday, December 06, 2011
1:00-3:00 PM
Eng 4201
Ahmed K. Alazzawe, PhD IT Candidate

Abstract

This dissertation presents a method of accurately determining the location of a person regardless of GPS access by using a multiple sensors that are readily available on smartphones (i.e. an accelerometer and a magnetometer, in conjunction with a camera). This location information can be used to develop a dynamically updating context-aware Smart Phone Book. The Smart Phone Book can deter unwanted calls, redirect calls to a more convenient form of communication, or transparently and automatically enable silent mode according to location. A required initial training phase allows the smartphone to recognize the user’s most frequented locations. This recognition is based on features derived from images of each environment that the user records with the phone’s camera and tags for the Smart Phone Book. Three feature-extracting methods—grayscale histogram, RGB component histogram, and RGB histograms—are explored comparatively in relation to the required storage space and effect on classification accuracy. Various classification algorithms in conjunction with their different parameter settings are analyzed for classification accuracy and speed. A novel method is introduced that takes advantage of sensors available on smartphones to further improve location detection accuracy to over 90 percent. A copy of this doctoral dissertation is on reserve at the Johnson Center Library.

Oral Defense of Doctoral Dissertation: Risk-based Models for Managing Data Privacy in Healthcare

Thursday, December 08, 2011
1:00-3:00 PM
Eng 3507
Ahmed A. Al-Faresi, PhD IT Candidate

Abstract

Current research in health care lacks a systematic investigation to identify and classify various sources of threats to information privacy when sharing health data. Identifying and classifying such threats would enable the development of effective information security risk monitoring and management policies. In this research I put the first step towards identifying and classifying privacy threats from a selection of health data exchange scenarios. Specifically I investigate data sharing scenarios that occur within a health care organization, between a health organization and a research group, and between patients and online social networks. I first derive the privacy requirements from legislative laws for protecting patient privacy in the U.S., namely the Health Insurance Portability and Accountability Act (HIPAA). Using the derived requirements I develop methods to enforce them in the data sharing scenarios specified. I use risk modeling to quantify the privacy threat in each sharing scenario and I incorporate that risk intelligence to develop security solutions to counteract the vulnerabilities found.

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