Title :
Mixed gradient based fast learning algorithm with variable learning gain and selective updates for layered neural nets
Author :
Xiang, Zengjun ; Bi, Guangguo
Author_Institution :
Dept. of Radio Eng., Southeast Univ., Nanjing, China
Abstract :
A new fast adaptive learning algorithm is put forward, which uses the steepest descent method combined with the conjugate gradient method. Line search unconstrained optimization is adopted to adjust adaptively the learning gain. Computer simulation results are illustrated
Keywords :
conjugate gradient methods; feedforward neural nets; learning (artificial intelligence); optimisation; conjugate gradient method; fast adaptive learning algorithm; layered neural nets; line search unconstrained optimisation; mixed gradient method; selective weight updates; steepest descent method; variable learning gain; Backpropagation algorithms; Bismuth; Computational complexity; Computer simulation; Convergence; Gain; Gradient methods; Neural networks; Nonhomogeneous media; Optimization methods;
Conference_Titel :
Singapore ICCS/ISITA '92. 'Communications on the Move'
Print_ISBN :
0-7803-0803-4
DOI :
10.1109/ICCS.1992.255023
