2CSE50E14: Pattern Recognition

Learning Outcomes: 
After learning the course the students should be able to
Understands the fundamental pattern recognition and machine learning theories
Able to design and implement certain important pattern recognition techniques
Able to apply the pattern recognition theories to applications of interest.
Distinguish supervised learning methods from the unsupervised ones.
Able to apply supervised learning methods (model-based maximum likelihood, k-nearest neighbours) to the classifier design.
Able to apply k-means clustering algorithm.
Syllabus: 
Unit NoTopics
Introduction

Paradigms for pattern recognition, Statistical and Syntactic  pattern recognition, Soft and Hard computing schemes for pattern recognition. Statistical Pattern Recognition- Patterns and classes, Supervised, Semi- supervised, and Unsupervised classification

Representation

Vector space representation of patterns and classes, patterns and classes as strings, Tree-based representations, Frequent itemsets for representing classes and clusters, Patterns and classes as logical formulas

Proximity Measures

Dissimilarity measures, metrics, similarity measures, Edit distance, Hausdorff metric between point sets, Kernel functions, Contextual and conceptual similarity between points

Dimensionality Reduction

Feature selection: Branch and bound, Sequential feature selection, Feature extraction: Fisher's linear discriminant, Principal components as features; Nearest Neighbor Classifiers- Nearest neighbor classifier, Soft nearest neighbor classifiers, Efficient algorithms for nearest neighbor classification, K-nearest neighbor classifier, minimal distance classifier, condensed nearest neighbor classifier and its modifications

Bayes Classifier

Bayes classifier, naïve Bayes classifier, Belief net; Decision Trees- Axis- parallel and oblique decision trees, Learning decision trees, Information gain and Impurity measures

Linear Discriminant Functions

Characterization of the decision boundary, Weight vector and bias, Learning the discriminant function, Perceptrons; Support Vector Machines- Maximizing the margin, Training support vector machines, Kernel functions

Clustering

Clustering process, Clustering algorithms, Clustering large datasets

Combination of Classifiers

AdaBoost for classification, Combination of homogeneous classifiers, Schemes for combining classifiers

Text Books: 
Name : 
Pattern Recognition: An Introduction
Author: 
Susheela D evi and M. Narasimha Murty
Publication: 
Universities Press, Hyderabad, 2011.
Reference Books: 
Name: 
Pattern Classification
Author: 
R. O. Duda, P. E. Hart and D. G. Stork
Publication: 
John Wiley and Sons, 2000
Syllabus PDF: 
AttachmentSize
PDF icon ELECTIVE I (PR) .pdf104.94 KB
branch: 
BDA
Course: 
2014
2016
Stream: 
B.Tech