ECE 8443
Instructor:
PATTERN RECOGNITION
Spring ‘06
N.H. Younan, Simrall 225
Phone: 325-3721, FAX: 325-2298
E-mail: younan@ece.msstate.edu
Course Information: www.ece.msstate.edu/~younan
Objective:
Provide a broad perspective of pattern recognition engineering approaches,
models, theoretical foundation, limitations, and practical constraints.
Prerequisites:
Probability and Random Processes
Required Text:
Pattern Classification - Duda, Hart, and Stork, second edition and Computer
Manual in MATLAB to Pattern Classification – Stork and Yom-Tov, Wiley
Interscience
References:
Pattern Recognition: Statistical, Structural, and Neural Approaches – Schalkoff
Pattern Recognition and Image Analysis – Gose, Jhonsonbraugh, and Jost
Introduction to Statistical Pattern Recognition - Fukunaga
Grading:
Mid-term
Final
Homework and Computer Assignments
Semester Project/Presentation
25%
30%
20%
25%
Computer Assignments:
Publication format:
Abstract
Introduction
Brief discussion of Theory
Analysis of Results
Conclusion/Summary
Computer code listing
Project/Presentation:
Proposal:
Presentation:
Project report:
Report format:
Due February 21, a weekly progress report is required afterward
Last 2 days of classes, April 27 and May 2
Due last day of classes, May 3
Similar to Computer Assignments
Notes: 1. Prior to proposal submission, a one page summary of a related journal article is
due every Thursday, starting Thursday, 26.
2. All assignments are due by 11:30 am on the due date – Late assignments will not
be accepted.
3. Electronic submission of summaries, computer assignments, and reports is
required.
COURSE OUTLINE:
I. INTRODUCTION TO PATTERN RECOGNITION
A. Machine Perception
B. Pattern Recognition Systems
C. Design Cycle
D. Learning Models
II. BAYESIAN and MAXIMUM LIKELIHOOD APPROACHES
A. Decision Theory
B. Bayes Classifiers and Discriminant Functions
C. Maximum Likelihood Classifiers
D. Error Rates, Probabilites, and bounds
III. DIMENSIONALITY PROBLEM
A. Problem of Dimensionality
B. PCA and LDA
C. EM (Expectation-Minimization)
D. HHM (Hidden Markov Models)
IV. NONPARAMETRIC METHODS
A. Parzen Window
B. Nearest Mean
C. Nearest Neighbor
V. LINEAR DISCRIMINANT FUNCTIONS
A. Linear and Generalized Linear Discriminant Functions
B. Relaxation Procedures
C. Separable and Nonseparable Cases
VI. MACHINE LEARNING
A. Classifier Superiority
B. Bias and Variance
C. Reasampling
D. Combination and Comparison of Classifiers
VII. UNSUPERVISED CLUSTERING
A.
B.
C.
D.
E.
Mixture Densities
ML and Bayesian Approaches
Clustering
Criterion Functions for Clustering
Hierarchical Clustering