Pattern Recognition I

Overview

Introduction and mathematical preliminaries:What is Pattern recognition; Applications and Examples - Clustering vs. Classification; Supervised vs. unsupervised - Relevant basics of Linear Algebra, vector spaces - Probability Theory basics - Basics of Estimation theory - Decision Boundaries, Decision region / Metric spaces/ distances - Mathematical Assignments;Classification:Bayes decision rule, Error probability - Examples - Normal Distribution - Linear Discriminant Function (equal covariance matrices) - Non-linear Decision Boundaries (unequal covariance matrices) - Mahalanobis Distance - K-NN Classifier - Fishers LDA - Single Layer Perceptron - Multi-layer Perceptron - Training set, test set; standardization and normalization;Clustering:Basics of Clustering; similarity / dissimilarity measures; clustering criteria - Different distance functions and similarity measures - Minimum within cluster distance criterion - K-means algorithm;Single linkage and complete linkage algorithms, MST - K-medoids, DBSCAN - Data sets - Visualization; Unique Clustering; No existence of clusters;Feature selection:Problem statement and Uses; Algorithms - Branch and bound algorithm, sequential forward / backward selection algorithms, (l,r) algorithm;Probabilistic separability based criterion functions, interclass distance based criterion functions;Feature Extraction:PCA + Kernel PCA;Recent advances in Pattern Recognition:Structural PR, SVMs, FCM, Soft-computing and Neuro-fuzzy techniques, and real-life examples