Title :
Learning algorithm for neural networks by solving nonlinear equations
Author :
Aoki, Kenichi ; Kanezashi, Masakazu ; Maeda, Chieko
Author_Institution :
Dept. of Manage. Inf., Hiroshima Prefectural Univ., Japan
Abstract :
The BP (backpropagation) process is a popular learning algorithm for neural networks. Despite of many successful applications, the BP process has some known drawbacks. These drawbacks stem from that the BP process is a gradient based optimization procedure without a linear search. In this paper, a new learning algorithm is presented based on a solution method of nonlinear equations. Compared with the former optimization procedure, the proposed method often converges faster to desired results. Newton´s method is basically applied to solve the nonlinear equations. However, the major difficulty with Newton´s method is that its convergence depends on an initial point. In order to assure a global convergence, independent of an initial point, the Homotopy continuation method is employed.
Keywords :
backpropagation; neural nets; nonlinear equations; optimisation; AI; Homotopy continuation method; backpropagation; convergence; global convergence; gradient based optimization procedure; learning algorithm; neural networks; nonlinear equations; Backpropagation algorithms; Cities and towns; Convergence; Informatics; Linear approximation; Multi-layer neural network; Neural networks; Newton method; Nonlinear equations; Optimization methods;
Conference_Titel :
Neural Networks to Power Systems, 1993. ANNPS '93., Proceedings of the Second International Forum on Applications of
Conference_Location :
Yokohama, Japan
Print_ISBN :
0-7803-1217-1
DOI :
10.1109/ANN.1993.264305