CS 700
Quantitative Methods & Experimental Design in Computer Science

Time/Location: Tuesday 4:30-7:10, Innovation Hall 208
Instructor: Jana Kosecka
Office hours:  2-3pm Wednesday
Contact: Office 4444 Research II, e-mail: kosecka@gmu.edu, 3-1876
Course web page: http://www.cs.gmu.edu/~kosecka/cs700/

The course covers treatment of the models and practices of experimental computer science and review of tools for applied science in general. The first half of the couse will introduce some elementary methods for statistical analysis of low dimensional (1D-2D) data, hypothesis testing, uses of analytic and simulation models, design of performance and quality metrics and interpretation and presentation of experimental results. In the second half we will use mathematical framework of linear algebra to discuss tools for data fitting, singular value decomposition, dimensionality reduction and pattern recognition, which are critical components for analyzing and statistically characterizing higher dimensional data.

Schedule, Homeworks, Handouts

Prerequisites:

STAT 344 (or equivalent), at least two 600 level courses in computer science, and doctoral status.

Materials:

The first half of the course will use the required textbook and the second half of the course will be based on on-line resources and lecture notes and handouts given by instructor. The computing environment will be MATLAB.

Required Textbook:

  • David Lilja, Measuring Computer Performance: A Practitioner's Guide, Cambridge University Press, 2005. ISBN: 05216-4670-7.
  • Online resources:

  • G. Strang's on-line Linear Algebra Course
  • Matlab
  • Statistical Pattern Recognition
  • Recommended Textbooks:

  • Raj Jain, The Art of Computer Systems Performance Analysis, John Wiley, 1991, ISBN: 0-471-50336-3. Please download the errata
  • P. Cohen, Empirical Methods for Artifical Inteligence, MIT Press, 1995.
  • Grading:

    Homeworks: 30 %
    Exams: Midterm 25%, Final 20%  
    Project: 25%
    Late policy: Each student will have a 3 day late submission budget, which could be used towards late submssion on the homeworks.

    Tentative List of Topics:
     

    Topics 
      Introduction to Experimental Techniques on CS
      Summarizing Measured Data, Comparing Alternatives
      Hypothesis testing, Characterization of errors
      ANOVA, Confidence Intervals
      Introduction to Simulation  
      Analytical modelling, Regression
      Multi-dimesional data, Dimensionality reduction
      Clustering, Introduction to Pattern Recognition