DocumentCode :
468984
Title :
Speech recognition based on cooperative particle swarm optimizer wavelet neural network
Author :
Chen, Li-wei ; Zhang, Ye
Author_Institution :
Harbin Inst. of Technol., Harbin
Volume :
2
fYear :
2007
fDate :
2-4 Nov. 2007
Firstpage :
716
Lastpage :
720
Abstract :
In BP wavelet neural network, the learning algorithm is BP algorithm, it is the stochastic gradient algorithm virtually, and it is local search algorithm, using this algorithm, the network may get into local minimum, the result of network training is dissatisfactory. In this paper, the cooperative Particle Swarm Optimizer algorithm CPSO) being used to train the parameters of the Wavelet Neural Network. The CPSO is a variant of the Particle Swarm Optimizer (PSO) that splits the problem vector; for example a neural network weight vector; across several swarms. This paper investigates the influence that the number of swarms used (also called the split factor) has on the training performance of a wavelet neural network. Then the CPSO-WNN being used in noise speech recognition, simulation results show compared with the BP network, the iterative number, error of the function approximation and the performance of the network are highly improved than BP network, the recognition rate are highly improve also.
Keywords :
backpropagation; gradient methods; neural nets; particle swarm optimisation; speech recognition; wavelet transforms; BP algorithm; cooperative particle swarm optimizer; local search algorithm; network training; speech recognition; stochastic gradient algorithm; wavelet neural network; Convergence; Feedforward neural networks; Feeds; Multi-layer neural network; Neural networks; Particle swarm optimization; Signal analysis; Speech recognition; Wavelet analysis; Wavelet transforms; Cooperative particle swarm optimizer; noise speech recognition; speech recognition; wavelet neural network;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Wavelet Analysis and Pattern Recognition, 2007. ICWAPR '07. International Conference on
Conference_Location :
Beijing
Print_ISBN :
978-1-4244-1065-1
Electronic_ISBN :
978-1-4244-1066-8
Type :
conf
DOI :
10.1109/ICWAPR.2007.4420762
Filename :
4420762
Link To Document :
بازگشت