Many subareas of computer science are inherently empirical in nature. Whether one is designing network routing algorithms, tuning database search parameters, or employing a machine learning algorithm to solve a robot control problem, there exist a number of common steps in the research process. These include: the proper construction of experimental questions, the design of methods to get at these questions, and the evaluation of the empirical results. In this graduate-level seminar, we will discuss the formulation of empirically-testable hypotheses as applied to different sub-fields of computer science and related areas, the design of experiments in order to test these hypotheses, and a range of statistical methods that are available for the evaluation and analysis of experimental results.
Topics will include:
Where: SEC A0133
When: T/Th 3:00-4:15
Prerequisites: Permission of the instructor. Fundamentally, we are looking for: a course in statistics, and an advanced course in an empirical computer science area (e.g., networks (CS 4133/5133/G5143/G6143), robotics (CS 4023/5023), operating systems (CS 4113/5113), machine learning (CS G5033), artificial intelligence (CS G4013), database management (CS G4513), computer architecture (CS G5633))
Last modified: Wed Sep 10 00:47:07 2008