New Jersey Institute of Technology
Department of Computer Science
CS782 - Pattern Recognition and Applications -
Spring'2003
Monday, 6:00 - 9:05 PM, KUPF 205
Course
Description | Readings | Tentative Contents | Grading
Policy | Miscellaneous
Chengjun Liu, Ph.D.
Phone: 973-596-5280
Email: chengjun.liu
njit.edu
Office: GITC
4306
Hours: MW 3:30PM-5:00PM or by appointment
Course Description
- Study of recent advances in development
of
(statistical and syntactic) pattern recognition algorithms,
approximation,
and estimation techniques. Topics include statistical estimation
theory,
classifier design, parameter estimation and unsupervised learning, bias
vs. variance, nonparametric techniques, linear discriminant functions,
tree
classifiers, feature extraction, and clustering. Additional topics
include
Support Vector Machines (SVM), Bayesian learning, Hidden Markov Models
(HMM),
evolutionary computation, neural networks, with applications to signal
interpretation,
time series prediction, and Biometrics.
Readings
- R.O. Duda, P.E. Hart, and D.G. Stork, Pattern
Classification, 2nd edition, John Wiley & Sons, 2001.
- K. Fukunaga, Introduction to Statistical Pattern Recognition,
2nd edition, Morgan Kaufmann, 1990.
- V. N. Vapnik, The Nature of Statistical Learning Theory,
2nd edition, Springer, 2000.
- T. M. Mitchell, Machine Learning, WCB/McGraw-Hill, 1997.
- S. Haykin, Neural Networks - A Comprehensive Foundation,
2nd edition, Prentice-Hall, 1999.
- Selected papers.
Tentative
Contents
- Introduction
- Pattern Recognition Fundamentals
- Formulation of Pattern Recognition Problems
- Process of Classifier Design
- Related Fields: AI, ML, NN, EC, SLT
- Bayes Decision Theory - Decision Rules
- The Bayes Decision Rule for Minimum Error
- The Bayes Decision Rule for Minimum Cost
- The Neyman-Pearson Decision Rule
- The Minimax Decision Rule
- Bayes Decision Theory - Error Bounds
- Error Probability in Hypothesis Testing
- Upper Bounds on the Bayes Error
- Parametric Classifier Design
- The Bayes Linear Classifier
- Linear Classifier Design
- Quadratic Classifier Design
- Piecewise Classifier Design
- Parameter Estimation
- Maximum-Likelihood Estimation
- Bayesian Estimation
- Bootstrap Methods
- Feature Extraction and Mapping for Representation
- Redundancy Reduction - Data Reduction, Dimensionality Reduction
- Linear Component Analysis
- Nonlinear Component Analysis - Kernel Methods
- Feature Extraction and Mapping for Classification
- Linear Discriminant Analysis
- Nonlinear Discriminant Analysis - Kernel Methods
- Statistical Learning Theory
- Structural Risk Minimization
- Support Vector Machines (SVM)
- Applications: Biometrics
- Face Detection
- Face Tracking
- Face Recognition
- Nonparametric Density Estimation
- Parzen Density Estimation
- KNN Density Estimation
- Expansion by Basis Functions
- Nonparametric Classifier Design
- Parzen Approach and its Error Estimation
- KNN Approach and its Error Estimation
- Clustering
- Parametric Clustering
- Nonparametric Clustering
- Machine Learning for
Pattern Recognition
- Bayesian Learning
- Neural networks
- Decision Trees
- Evolutionary Computation for
Pattern Recognition
- Genetic Algorithms (GA)
- Evolutionary Strategy (ES)
- Evolutionary Programming (EP)
Grading
Policy
Projects (topics are related
to our course Contents)
Presentation (~10 mins.)
Paper (~10 pages)
Class attendance
Miscellaneous