Course Number: STAT 59800 Section SK1 (47822)
Lectures: Monday, Wednesday, and Friday, 9:30am-10:20pm at 103 University Hall
Textbook: Probabilistic Graphical Models: Principles and Techniques, by Daphne Koller and Nir Friedman, The MIT Press, 2009 (required)
Course Webpage: http://learning.stat.purdue.edu/courses/598l/start
Course Mailing List: email@example.com
Instructor: Sergey Kirshner Email: skirshne at purdue dot edu
Office Hours: Mondays, 1:30-2:30pm in HAAS 118 or by appointment
Probabilistic graphical models provide a convenient framework for modeling of joint distributions by utilizing graphs to represent the dependence among the variables. The course introduces several such frameworks, including Bayesian (belief) networks, Markov random elds, and covers topics related to representation, exact and approximate inference, and parameter and structure estimation in models for high-dimensional data.
A basic course on probability (e.g., STAT 516/STAT 519) and some programming experience (e.g., STAT 598G) is required; a course in linear algebra is recommended; a course in machine learning (e.g., STAT598A/STAT598N/CS578) would be helpful but is not required. Prerequisites can be waived with a consent of instructor.
This is a graduate level course. The lectures will provide you with fundamentals of the covered topics, and will move fairly fast. While the intuition and the important details will be emphasized as well, it would be your responsibility to ll some of the gaps either from the textbook or from the literature mentioned in class. Programming assignments and the project while extremely useful and hopefully enjoyable will also require a significant time commitment. Depending on your preparation and programming experience, expect to devote 10-15 hours per week to this course outside of lecture.
On the flip side, you have a say on the matter of what topics are covered as the last part of the course can be tailored to your interests. You will also likely have a significant face time with the instructor, especially when you are designing your project. One official office hour per week is misleading; during the last offering of the course, students frequently met with the instructor using appointments.
Essential course material will be presented in lectures over roughly the first two-thirds of the semester. The remaining time will be devoted to additional topics (presented by the instructor and by the students) and to student project presentations.
Your performance will be evaluated based on 4-6 homeworks (40%), a 10-week course project (40%), and course participation (20%).
The homeworks will be a combination of theoretical exercises and programming assignments consisting of implementation of the algorithms discussed in class.
The course project will be performed either individually or in groups of two (depending on the class size); each group is required to make a presentation on its project during the last week of classes and to submit its final project report during the week of finals.
This course is an elective taught in a small-size class setting. Your active class participation, be it questions, ideas, comments, suggestions, is not only encouraged, it is essential. In addition to the project, each student is expected to make a presentation in the last third of the course on an additional graphical models topic of their choice (either a brief survey or a short presentation on a seminal idea).
You are encouraged to discuss the material with your classmates outside of lectures to enhance your understanding of the subject. You are also allowed to discuss the homework assignments with your classmates as long as no written (or typed) notes are taken. However, your submitted homework assignments should be your own work.
Written portions of homeworks will be collected at the beginning of class on their due dates; they can also be emailed to the instructor. Submission guidelines for programming assignments will be discussed in class. No late homeworks will be accepted.
For the project, you will be able to choose from among the topics suggested by the instructor to come up with your own with instructor's approval. Your are encouraged to discuss your projects with the instructor. If done by a group, the proposed work has to be clearly divided between the participants, and the project report should outline the contribution of each participant.
In the event of a major campus emergency, course requirements, deadlines, and grading percentages (points) are subject to changes that may be necessitated by a revised semester calendar or other circumstances beyond the instructor's control. The information about the changes in this course will be announced on the course's webpage, over email, and in class (whenever possible).
Cheating and plagiarizing will not be tolerated. If caught, your punishment may range from a score of zero on an assignment to a failing grade in the course with a referral to the University disciplinary committee. (See regulations for student conduct) Please don't resort to cheating; if you are having trouble in the course, please talk to the instructor (me), and I may be able to direct you to additional resources.
I hope you enjoy the course. If you have comments or suggestions, I want to hear them. Please drop me a line or just stop by.