CS 228: Structured Probabilistic Models: Principles and Techniques

CS 228
Structured Probabilistic Models: Principles and Techniques
Winter 2010

Announcements

Course Information

228	Lecture:	TueThu 11:00-12:15PM, 3 units
	Location:	Gates B01

Useful Information and Handouts

Teaching Staff & Office Hours

Syllabus

Assignment Schedule

Handouts, Course Notes, Problem Sets

Class Wiki (Book Corrections and Partner Matching)

FAQ - problem set clarifications and other frequently asked questions

Course Description

The course covers modeling (knowledge representation) languages suitable for dealing with an uncertain world. The focus is on probabilistic models, including Bayesian networks, Markov networks, hidden Markov models, and influence diagrams. The course will cover syntax, probabilistic semantics, inference, and learning algorithms for these frameworks. It will also discuss recent applications to domains such as speech recognition, medical diagnosis, data mining, statistical text modeling, and robot motion planning.

Prerequisites

Familiarity with the basic concepts of probability theory. (Stat116 is sufficient but not necessary.)
Knowledge of basic computer science principles and skills at the level of CS103.
Mathematical ability and the ability to understand and analyze fairly complicated algorithms and data structures. (CS161 is sufficient but not necessary.)

Grading

CS228 will have nine weekly quizzes, three programming assignments, and two problem sets. The programming assignments should be relatively short, about 30 to 50 lines of Matlab code. Problem sets must be handed in to the submission box at the bottom of the Gates A wing (West) stairwell by the beginning of class on the day they are due. Programming assignments must be submitted electronically by 11:59 PM on the due date.

Recognizing that students may face unusual circumstances and require some flexibility in the course of the quarter, each student will have a total of seven free late (calendar) days to use as s/he sees fit. Once these late days are exhausted, any homework turned in late will be penalized at the rate of 20% per late day (or fraction thereof). Under no circumstances will a homework be accepted more than five days after its due date. Furthermore, there will be a maximum of two late days allowed for the last programming assignment, and no late days allowed for the last problem set. Late days are in 24 hour increments after the due date/time. Late homeworks should be turned in to the submission box. You must write the time and date of submission on the assignment. It is an honor code violation to write down the wrong time.

Programming assigments may be completed in teams of up to 2 students, will be implemented in MATLAB, and will typically require around 30-50 lines of coding. We will provide starter code and stubbed out functions that you will be implementing. The actual coding will tend to be relatively straightforward, so even students with little MATLAB experience should be able to complete the assignments. The third programming assignment will include an open-ended component that will allow you to implement a range of strategies and experiments, though we will provide a list of possibilities to help guide you. Late days also apply to programming assignments, and will be applied to everyone in the team. If you turn in a programming assignment one day late and you have no late days left, you will be penalized 20% while your partner will be deducted one late day.

There will be nine quizzes, which will be available online and due by Monday night of each week. The quizzes are designed to keep all students up with the material, allowing lectures to focus on the more complex issues. The quizzes are open book, but each student must take the quiz alone. Each student's top eight (highest scoring) quizzes will count in the grade.

Quizzes will count for 32% of the final grade (4% each), problem sets will count for 34%, and programming assignments will count for 34% (the third assignment will be worth more than the first two and will have some extra credit associated with it). Some extra credit may be awarded for class participation.

CS228 may be taken pass/no-credit. The requirements are as specified in the university regulations. You take the class, and get a letter grade as per the standard grading curve. Any letter grade higher than a C- is a passing grade, otherwise not. As a side note, we recommend at least attempting most of the work, since much of the learning occurs through solving homework problems.

Collaboration and Honor Code

Students may discuss and work on problem sets in groups but must write up their own solutions. When writing up the solutions, students should write the names of people with whom they dicussed the assignment. Also, when writing up the solutions students should not use notes from group work. If we start receiving solution sets which are too similar, we may reconsider this policy. Collaboration is not permitted on the quizzes. Programming assignments may be completed in teams of 2 students. Only one student needs to submit the code, but please indicate clearly in the submission README who was on the team.

Important Note: As we often reuse problem set questions from previous years, we expect the students not to copy, refer to, or look at the solutions in preparing their answers. It will be considered an honor code violation to intentionally refer to previous year's solutions. The purpose of problem sets in this class is to help you think about the material, not just give us the right answers. Moreover, some of the problems are taken from recently published papers. Students are expected to come up with the answers on their own, rather than extracting them from published solutions. Therefore, please restrict attention to the books mentioned on the webpage when solving problems on the problem set. If you do happen to use other material, it must be acknowledged clearly with a citation on the submitted solution.

Communication

We strongly encourage students to come to office hours. If that is not possible, the questions should be sent to the staff list (cs228-qa@cs.stanford.edu), not to the individual TAs and not to the general cs228 mailing list. If a homework clarification is posted after a student has completed an assignment, the student should contact us as soon as possible to check if the assumptions s/he made are going to be accepted.

As explained above, late homeworks should be turned in to the submission box at the bottom of the Gates A wing (West) stairwell.

Please do not e-mail TAs with grading questions. If you want us to explain why we took points off, you can talk to us after class or during office hours. If you want a regrade, please write an explanation and hand the homework and the explanation to the TAs during office hours or after class.

Occasionally we may need to broadcast a message to entire class. When you sign up on Axess, you will automatically be subscribed to the CS228 Mailing List.

Textbook Information

The primary reading materials will be the textbook by Professor Koller, co-authored with Nir Friedman. The book, finally published this past fall, is available at the Stanford Bookstore.

Optional books containing relevant material:

Robert G. Cowell, A. Philip Dawid, Steffen L. Lauritzen, David J. Spiegelhalter. Probabilistic Networks and Expert Systems Springer-Verlag, 1999.
Judea Pearl. Probabilistic Reasoning in Intelligent Systems Morgan Kaufmann, 1988.

Comments to CS228 Staff