CmpE 621 Pattern Recognition

Fall Semester 2000-2001

Ethem Alpaydin

Topics covered by lectures

Week 1 Introduction
Week 2 Statistics Review: Sampling, Parameter Estimation.
Week 3 Statistics Review Cont'd: Interval Estimates, Hypothesis Testing, Regression, Analysis of Variance. Bayesian Classification.
Week 4 Multivariate Analysis Review. Parametric Classification. 1st HW on Polynomial Regression, due Nov 2
Week 5 Parametric Discrimination, Principal Components Analysis, Logistic Discrimination. 2nd HW on Parametric Discrimination and PCA.
Week 6 Fisher's Linear Discriminant, Discrimination by Regression.
Week 7 Nonparametric Methods for Probability Density Estimation and Classification Kernel-based methods, k-nearest neighbor. 3rd HW on Fisher's LD, Logistic Discrimination and Discrimination by Regression.
Week 8 Nonparametric methods for Regression. Generalized Additive Models.
Week 9 Unsupervised Learning. k-means. Gaussian mixtures and Expectation-Maximization Algorithm.4th HW on Nonparametric Classification
Week 10 Hidden Markov Models. Baum-Welch algorithm.
Week 11 Decision Trees. ID Trees, multivariate splits, error correcting output codes.
Week 12 Review of the course.

Catalog Data

Bayes Decision Theory. Parametric and Nonparametric Methods. Linear Discriminant Functions. Higher Order Discriminants with Emphasis on Artificial Neural Network Based Learning Methods. Unsupervised Learning and Clustering. Case study: Vision.

Reference Books

Duda, R., Hart, P. (1973) Pattern Classification and Scene Analysis, Wiley.

Fukunaga, K. (1990) Introduction to Statistical Pattern Recognition, 2nd Edition, Academic Press.

McLachlan, G. (1992) Discriminant Analysis and Statistical Pattern Recognition, Wiley.

Schalkoff, R. (1992) Pattern Recognition: Statistical, Structural, and Neural Approaches, Wiley.

Instructor

Dr Ethem Alpaydin, Associate Professor. Department of Computer Engineering, Bogazici University alpaydin@boun.edu.tr

Goals

To introduce the student to the problems related to pattern recognition and discuss how solutions may be attempted using statistical techniques. This course is followed by and is a prerequisite for CmpE 545 Artificial Neural Networks.

Prerequisite by Topic

Undergraduate level calculus, probability theory. Prior experience in a high-level programming language.

Topics

Introduction to Pattern Recognition

Probability Review

Statistics Review

Multivariate Analysis

Bayes Decision Theory

Parametric Techniques

Nonparametric Techniques

Unsupervised Learning and Clustering

Computer Usage

Almost all homeworks require computer simulations.

Grading

1 Project 0.30

1 Final 0.30

Homeworks 0.40