Title :
Use of dynamic tunneling with backpropagation in training feedforward neural networks
Author :
Chowdhury, Pinaki Roy ; Singh, Y.P. ; Chansarkar, R.A.
Author_Institution :
Defence Terrain Res. Lab., Delhi, India
Abstract :
An algorithm is proposed for supervised training in multilayer feedforward neural networks. The proposed algorithm consists of a gradient descent technique with a dynamic tunneling method for global optimization. Relative to the gradient descent technique like the backpropagation algorithm, the proposed technique gives a faster convergence, and the convergence properties of the proposed algorithm is demonstrated through three problems
Keywords :
backpropagation; convergence of numerical methods; feedforward neural nets; gradient methods; optimisation; backpropagation algorithm; convergence properties; dynamic tunneling; feedforward neural networks training; global optimization; gradient descent technique; multilayer feedforward neural networks; supervised training; Backpropagation algorithms; Containers; Convergence; Feedforward neural networks; Intelligent networks; Multi-layer neural network; Neural networks; Optimization methods; Quantum computing; Tunneling;
Conference_Titel :
TENCON '98. 1998 IEEE Region 10 International Conference on Global Connectivity in Energy, Computer, Communication and Control
Conference_Location :
New Delhi
Print_ISBN :
0-7803-4886-9
DOI :
10.1109/TENCON.1998.797057
