09AE 6122 Pattern Recognition and Analysis

Dear M.Tech Signal Processing students,

This is the course page for the course 09AE6122 pattern Recognition and Analysis

COURSE PLAN

Course No:09AE6122 Title: PATTERN RECOGNITION AND ANALYSIS

(L-T-P): 3-0-0 Credits : 3

Module I: Introduction - features, feature vectors and classifiers,

Supervised versus unsupervised pattern recognition. Classifiers based

on Bayes Decision theory- introduction, discriminant functions and

decision surfaces, Bayesian classification for normal distributions,

Estimation of unknown probability density functions, the nearest

neighbour rule. Linear classifiers,- Linear discriminant functions and

decision hyper planes, The perceptron algorithm, MSE estimation,

Logistic determination, Support Vector machines.

Contact Hrs :11, % Marks in sem exam= 25

Module II: Non-Linear classifiers- Two layer and three layer

perceptrons, Back propagation algorithm, Networks with Weight

sharing, Polynomial classifiers, Radial Basis function networks,

Contact Hrs :5, % Marks in sem exam= 13

FIRST INTERNAL TEST

Support Vector machines- nonlinear case, Decision trees, combining

classifiers, Feature selection, Receiver Operating Characteristics

(ROC) curve, Class separability measures, Optimal feature

generation, The Bayesian information criterion.

Contact Hrs :5, % Marks in sem exam= 12

Module III: Feature Generation 1- Linear transforms- KLT, SVD,

ICA, DFT, DCT, DST, Hadamard Transform, Wavelet Transform,

Wavelet Packets etc- Two dimensional generalizations -

Applications. Feature Generation 2- regional features, features for

shape and characterization, Fractals, typical features for speech and

audio classification, Template Matching, Context dependent

classification-Bayes classification, Markov chain models, HMM,

Viterbi Algorithm. System evaluation - Error counting approach,

Exploiting the finite size of the data.

Contact Hrs :11, % Marks in sem exam= 25

SECOND INTERNAL TEST

Module IV: Clustering - Cluster analysis, Proximity measures,

Clustering Algorithms - Sequential algorithms, Neural Network

implementation. Hierarchical algorithms - Agglomerative algorithms,

Divisive algorithms. Schemes - based on function optimization -

Fuzzy clustering algorithms, Probabilistic clustering, K - means

algorithm. Clustering algorithms based on graph theory - Competitive

learning algorithms, Binary Morphology Clustering Algorithms

Boundary detection methods, Valley seeking clustering, Kernel

clustering methods. Clustering validity.

Contact Hrs :10 , % Marks in sem exam=25

END SEMESTER EXAMINATION