Princeton University
Computer Science Department

Computer Science 598A
Boosting: Foundations & Algorithms

Rob Schapire

Spring 2012


Directory
General Information | Schedule & Readings | Assignmentsblackboard

Course summary
Staff, lectures, problem sessions, office hours
Prerequisites
Textbook
Grading and workload
Collaboration and outside help
Auditors
Breadth requirement for CS graduate students
Communication


Course summary

This course will provide an in-depth examination of the boosting approach to machine learning, with emphasis on theoretical foundations.  Likely topics include some or all of the following:


Staff, lectures, problem sessions, office hours

Lectures: 

TTh, 11-12:20, Computer Science Building, room 102.

Problem sessions:

F, 11-11:50, Friend Center, room 110.  (First meeting will be on Friday, Feb. 17.)

These problem sessions (what the registrar calls "precepts") are intended for discussing and answering questions about the homeworks for the preceding week.

Professor:

Rob Schapire:  407 CS Building, x8-7726, schapire@cs
Office hours: by appointment.  To schedule an appointment, visit wass.princeton.edu, or send me email if no convenient times are available.

 


Prerequisites

COS 402 or 424 or 511, or permission of instructor.

I will also assume general familiarity with elementary notions from probability, and to a lesser extent, calculus and linear algebra.  Students should also be prepared to do rigorous mathematics, including proofs.


Textbook

Boosting: Foundations and Algorithms
by Robert E. Schapire and Yoav Freund
MIT Press, 2012

Unfortunately, this book will not actually be released by the publisher until May (or possibly slightly later, if production falls behind schedule).  We have instead made arrangements to provide photocopies of page proofs of the book.  You have two options:

  1. You can purchase only a photocopy of the sections that we are covering in the course for $25.

  2. Alternatively, you can both purchase a photocopy for use during the semester, and also pre-order a copy of the book, which will be delivered to you as soon as it is available.  The total cost for both the photocopy and the pre-ordered book will be $55 (thanks to a special discount that the publisher is providing).

For either option, to get your copy, you will need to bring payment to Pam DelOrefice in room 212 of the Computer Science Building.  Make your check payable to "Trustees of Princeton University."  You can also pay by cash, but only if you have the exact amount.  Note that Pam is only in her office Monday through Thursday until 2:15pm, and is not in her office at all on Fridays.  When she is not available, you can instead go to Nicole Wagenblast in room 220.


Grading and workload

Regular homeworks will be assigned on a weekly basis.  These will generally consist of paper-and-pencil exercises from the textbook.  There is no programming in this course.  Homeworks are due at the beginning of the Friday problem session, and will not be accepted late (except in emergency situations).  Homeworks will not be graded, but will only be checked to be sure that an attempt was made to solve each problem.  This means that homeworks do not have to be written up nicely.  It is okay, for instance, to turn in whatever scribbles you made in attempting each problem, provided that it is clear and evident that a genuine attempt was made on each one.

There will be two midterms, one in class (or possibly the problem session), and one take-home.  There will also be a final exam.

Final grade will be based roughly on:  midterms (26% each), final exam (36%), homeworks (12%).  In addition, final grades may be adjusted upward for regular and positive class participation, or downward for conspicuous absence from lecture or problem sessions.


Collaboration and outside help

You are welcome to work with other students in the course on the assigned exercises.  However, you should only turn in your own work and writing.  If you worked with others, please indicate this on your homework.

Please do not search the internet, library, etc. for answers to homework problems.  This really defeats the purpose of the homeworks, which is to give you practice solving problems in this area of study.  Try to solve them by yourself, by looking over the book or lecture notes, or by discussing them with other classmates (who have not already solved them).  If you are stuck or confused, you certainly can ask me for help.

Consulting any kind of website, blog, forum, mailing list or other service for discussing, purchasing, downloading or otherwise obtaining or sharing homework solutions is strictly forbidden (except for the class mailing list, as discussed below).


Auditors

Auditors are welcome, and are encouraged to take part in class discussions.  If you wish to receive official "credit" for auditing this course, you must attend the vast majority of the lectures.


Breadth requirement for CS graduate students

If you are a CS graduate student, you can take this course for credit in partial fulfillment of the department's breadth requirement.  However, for this course, you cannot pass this requirement by only taking the final exam.  Students who get an A- in the course are assured of passing the requirement; a lower grade may also be sufficient, as will be determined on a case-by-case basis at the end of the semester.


Communication

An email list is maintained by the university for all students who are registered for the course.  This list will be used for important general announcements, such as last minute corrections to the homeworks, changes in due date, etc.  To avoid missing these announcements, it is imperative that you be registered for the course.

This list can also be used by students for discussing course material, general-interest questions about the homeworks, etc.  Of course, if your question is specific to your own work, you will probably want to contact me directly, and you should always avoid giving away the solution to a problem in a general posting.

You can post to the list either using blackboard, or by sending email to COS598A_S2012@Princeton.EDU from your regular Princeton email account.