EE E6887: Statistical Pattern Recognition

EE E6887: Statistical Pattern Recognition, Fall 2005

COURSE BENEFITS:

Gain knowledge and practical experience in statistical pattern recognition
Learn practical skills and analytic background for building pattern recognition applications

PROFESSOR Shih-Fu Chang:

Extensive experience in digital video and multimedia
NSF CAREER Award, Office of Naval Research YIA award, IBM Faculty Research Award, Best Paper Awards from ACM, IEEE, and SPIE
Director of Columbia's Digital Video and Multimedia Lab

APPLICABLE DEGREE PROGRAMS:

Recommended course for EE MS concentrations in New Media Engineering and Telecommunications Engineering
Recommended course for the Signal Processing Track in the new EE undergraduate curriculum
Most courses 4000-level and above can be credited to all degree programs.

Lecturer/Manager:	Professor Shih-Fu Chang
Office Hours:	Mondays 2:30-3:30pm, CEPSR 709
Office Phone:	(212) 854-6894
Email Address:	[email protected]
Day & Time Class Meets on Campus:	Monday/Wednesday, 1:10-2:25 PM
Location:	CEPSR 415
Class Homepage:	http://www.ee.columbia.edu/~sfchang/course/spr
Credits for Course:	3
Class Type:	Lecture
Prerequisites:	Probability, Linear Algebra, or equivalent required Familiarity with Digital Image Processing Intended for beginning graduate students.
Description:	Introduction to theories, algorithms, and practical solutions of statistical pattern recognition. Topics covered include feature extraction, feature selection, Bayesian classifiers, neural networks, discriminative classifiers, clustering, performance evaluation, and fusion of models. Students will gain understanding of algorithm design, mathematical tools, and practical implementations of various applications, with special emphasis on image classification and multimedia indexing. Benchmark datasets, such as TRECVID news video, and evaluation metrics will be used for course projects.
Required Text(s):	Pattern Classification, 2nd Edition, Richard O. Duda, Peter E. Hart, and David G. Stork ISBN: 0-471-05669-3, 2000, Wiley
Reference Text(s):	The Elements of Statistical Learning, Trevor Hastie, Robert Tibshirani & Jerome Friedman, 2001, Springer Verlag. Machine Learning, Tom Mitchell, 1997 McGraw Hill.
Homework(s):	about 4 homeworks covering analytical questions and mini programming in language of your choice. Use of Matlab is suggested.
Project(s):	one
Paper(s):	None
Midterm Exam(s):	One, open book
Final Exam:	One, open book
Grading:	4 homeworks (30%), midterm (25%), final (25%), one project (20%)
Hardware requirements:	PC or notebook computer with access to Columbia’s systems
Software requirements:	Matlab is the recommended tool for the class. Software examples will be shown in class. Several public-domain tool libraries are available for running experiments. Matlab will be installed in the computer lab located in Mudd Rm. 251. Most students also find it convenient to purchase a student edition of Matlab for their own computers. However, students may choose any language of his/her choice for homework submission. All sample program and test data will be distributed on the course web site. There will be an online bulletin board for the class to exchange information and discuss common issues. Students need to have Columbia account in order to access the bulletin board.

Schedule for EE E6887: Statistical Pattern Recognition, Fall 2005
Class Date	Class No.	Topics/Chapters Covered	Homework Assigned	Due
	1
	2
	3
	4
	5
	6
	7
	8
	9
	10
	11
	12