CSE515 Statistical Methods in Computer Science – Spring 2013

Course description

Probabilistic graphical models have been applied to a wide ranging set of applications in computer science – such as computational biology, natural language processing, computer vision, robotics – enabling efficient inference, learning parameters/structures, and decision-making in problems involving a large number of variables and a vast amount of data. This course will provide you with a strong foundation on probabilistic graphical models such that you can apply graphical models to your own research problems or can address core research problems in graphical models. It covers largely four topics: representation of probabilistic graphical models such as Bayesian and Markov networks; probabilistic inference algorithms; learning algorithms for both parameters and structure of graphical models; and special graphical models and applications.

Meeting times and locations

  • Lectures: MW 10:30-11:50am @ MOR 221

Textbook

  • Required: Daphne Koller and Nir Friedman, Probabilistic Graphical Models: Principles and Techniques, MIT Press (http://pgm.stanford.edu/)

Pre-requisites

Students are required to have successfully completed STAT 341, 391, or an equivalent class, and are expected to have working knowledge of probability, statistics and algorithms. However, the class has been designed in a way that allows students with a strong computational background to catch up and participate. If you are not sure whether you have the necessary background knowledge, talk to the instructors.

Online resources

Access is currently limited to students enrolled in the course. Contact us if you need assistance.

Grading

  • 3 Homework assignments (15% each)

  • 2 Midterms (15% each)

  • 1 Project (35%)

  • Class participation (5%)

Only the top 4 scores from the 3 HWs and 2 midterms will be used: this way you can mess up on one of them and still get a perfect score. Homeworks will have a programming component in Matlab.

Homework policy

Collaboration policy

HW assignments are expected to be done individually unless otherwise specified; each student must hand in their own answers. In addition, each student must write their own code in the programming part of the assignments. You may discuss the subject matter with other students in the class, but all final answers must be your own work without referring to written notes from the discussion sessions. You also must indicate on each homework with whom you collaborated. You are expected to maintain the utmost level of academic integrity in the course.

Late policy

HW assignments are due at the beginning of class, unless otherwise specified. Students are allowed to use 3 total late days without penalty for the entire quarter. Each late day corresponds to 24 hours or part thereof. Once those days are used, you will be penalized by 10% of the maximum grade per day.

Regrade policy

If you feel that there is an error in grading your homework or midterm, please submit a written explanation to the TA, and we will consider your request to regrade.