MACHINE LEARNING January 1, 2013
COMS4771-001
COURSE INFO
Time
& Location |
Tue / Thu 1:10pm-2:25pm at HAMILTON HAL 702
|
Instructors
|
Adrian Weller,
adrian(at)cs(dot)columbia(dot)edu & Ilia Vovsha,
iv2121(at)columbia(dot)edu
|
Office
Hours |
Tue & Thu 2:30-3:15pm at CEPSR 6LE5 (Adrian)
|
TAs |
Xu Tan,
ttanxu(at)gmail(dot)com Peng Jiang,
pj2243(at)columbia(dot)edu Ran Yu,
ry2239(at)columbia(dot)edu
|
Bulletin
Board |
Available via courseworks.columbia.edu and is the best
|
Prerequisites:
Background in calculus, linear algebra, and statistics.
Programming ability in some (any) language.
Description: The course introduces various topics in machine learning. Material will include: Baeysian inference & decision theory, Gaussian and exponential family distributions, maximum likelihood, least squares, linear regression, linear classification, neural networks, statistical learning theory, support vector machines, kernel methods, mixture models, the EM algorithm, graphical models, and hidden Markov models. Students are expected to implement several algorithms in Matlab, and have some background in calculus, linear algebra, and statistics.
Recommended
Texts:
The following three
books are highly recommended.
Still
unpublished. Available online (password-protected) on class home page. Christopher M. Bishop, Pattern Recognition
and Machine Learning, Springer. 2006
First Edition is preferred. ISBN: 0387310738. 2006.
Optional
Texts: Available
at library (additional handouts and pointers to useful sites will also be provided).
Graded Work:
Grades will be based on homeworks (40%), the midterm (around 25%), and the final exam (around 35%). Any material covered in assigned readings, handouts, homeworks, solutions, or lectures may appear in exams. Your worst homework will not count towards your grade. If you miss the midterm and don't have an official reason, you will get 0 on it. If you have an official reason, your midterm grade will be based on the final exam. Tentative Schedule:
Date
Topic
January 22
Lecture 01: Introduction
January 24
Lecture 02: Basic Statistics
January 29
Lecture 03: Parametric Statistical Inference
January 31
Lecture 04: Parametric Statistical Inference
February 5
Lecture 05: Cross Validation & Parametric Paradigm
February 7
Lecture 06: Perceptron
February 12
Lecture 07: Neural Networks & BackProp
February 14
Lecture 08: Statistical Learning Theory (intro)
February 19
Lecture 09: Statistical Learning Theory (capacity)
February 21
Lecture 10: Statistical Learning Theory (bounds)
February 26
Lecture 11: VC Dimension
February 28
Lecture 12: Support Vector Machines
March 5
Lecture 13: Kernels
March 7
Lecture 14: Dimensionality Reduction
March 12
Lecture 15: Clustering
March 14
MIDTERM EXAM
March 19
Spring Recess (NO CLASS)
March 21
Spring Recess (NO CLASS)
March 26
Lecture 16: Mixtures of Gaussians, Latent variables, EM intro
March 28
Lecture 17: EM in more details
April 2
Lecture 18: Graphical Models...
April 4
Lecture :
April 9
Lecture :
April 11
Lecture :
April 16
Lecture :
April 18
Lecture :
April 23
Lecture :
April 25
Lecture :
April 30
Lecture :
May 2
Lecture : Class
Attendance: You
are responsible for all material presented in the class lectures,
recitations, and so forth. Some material will diverge from the textbooks so
regular attendance is important. Late
Policy: If you
hand in late work without approval of the instructor or TAs, you will
receive zero credit. Homework is due at the beginning of class on the due date. Cooperation
on Homework:
This
policy will be strictly enforced.
Discussion of Course Material:
See note at top of this page on the Bulletin Board.
Web
Page: The class
URL is: http://www.cs.columbia.edu/~coms4771
and will
contain copies of class notes, news updates and other information. Computer
Accounts: You
will need an ACIS computer account for email, use of Matlab
(Windows, Unix or Mac version) and so forth.
Specifically, the Duda & Hart text is a very gentle
introduction to many of the topics that will be covered in the first part of the course.
The Bishop book is a slightly more advanced discussion of many topics in machine learning.
The Jordan & Bishop text is very good on graphical models, which will be covered in the second half of the course.
Michael I. Jordan and Christopher M. Bishop, Introduction to Graphical Models.
R.O. Duda, P.E. Hart and D.G. Stork, Pattern
Classification, John Wiley & Sons, 2001.
V. Vapnik, Statistical Learning Theory, Wiley-Interscience, 1998.
Trevor Hastie, Robert Tibshirani and Jerome Friedman, The Elements of Statistical Learning,
Springer-Verlag New York USA, 2009. 2nd Edition. ISBN 0387848576.
D. Mackay, Information Theory, Inference and Learning Algorithms,
Cambridge University Press, 2003, available to download online.
You are encouraged to discuss HW problems with each other in small groups (2-3 people),
but you must list your discussion partners on your submission.
Solutions (code) must be written independently, sharing or copying of solutions is not allowed.
Of course, no cooperation is allowed during exams.
We have many interesting topics to cover, and many of you will have good questions.
Please try to post questions or ideas to the bulletin board on Courseworks so that everyone can participate.