III B.E. (Computer Engg.)

VII SEMESTER

7CP6.1 NEURAL NETWORKS

1.               INTRODUCTION TO AKm1CIAL NEURAL NETWORKS: Elementary Neuro physiology, Neural circuits for computations and becoming artificial resource as processing elements.

2.             BACK PROPAGATION: Back propagation network (BPN) approach and operation. Generalized data rule-updates of output layer weights and hidden layer weights. BPN implementation issue. Training data, network sizing, weights and learning parameters. BPN Application -Data compression and Paint quality inspection. Back propagation simulation for signal propagation -BPN data structure, signal propagation algorithms and error propagation.

3.               NEURAL NETWORK MEMORY: Introduction to Associative memory -Hamming distance, linear associative, Bi-directional Associative memory (BAM) Architecture, Processing, Mathematics and Energy Function. Hop field Memory -Discrete Hop field Memory. Continuous Hop field Model Traveling sales person problem. BAM simulation-Bi-directional connections, data structures, initialization algorithms and signal propagation.

4.            SIMULATED ANNEALING: Information theory and statistical mechanics concepts, Real and simulated Annealing. Boltzmann machine -Basic Architecture and processing, learning in Boltzmann machine and ,its practical consideration. Boltzmann simulator-Modified Boltzmann Networks its data structure and algorithm.

5.            COUNTER PROPAGATION NETWORK (CPN): Counter propagation Network Building Blocks -Input Layer, Instar, competitive Networks and 01ltstar. SPN Data Processing –Forward mapping, Training q>N and its complete implementation the CPN simulator -Data structure, Algorithms and complete simulator.

6.               SELF-ORGANIZING MAPS (SOM) -SOM data Processing, Data structure and learning algorithms.

Recommended Books:

1. James A. Freeman -Neural Networks Algorithms Applications and Programming Techniques, Pearson Education Asia.

2.      Simon Haykin -Neural Networks 2/e, Pearson Education Asia.

3.      Yagya Nard yan -Artificial Neural Networks, Prentice Hall India, 1999.