Title :
A novel nonlinear synapse neuron model guaranteeing a global minimum-Wavelet Neuron
Author :
Yamakawa, Takeshi
Author_Institution :
Kyushu Inst. of Technol., Iizuka, Japan
Abstract :
A novel nonlinear synapse neuron model named Wavelet Neuron is proposed, which realizes effective establishment of approximate wavelet expansion by learning instead of analytical procedure. The distinctive features of the model are guarantee of a global minimum and three decades higher speed of learning than feedforward multilayer neural networks
Keywords :
neural nets; wavelet transforms; Wavelet Neuron; learning; neuron model; nonlinear synapse; speed of learning; wavelet expansion; Neurons;
Conference_Titel :
Multiple-Valued Logic, 1998. Proceedings. 1998 28th IEEE International Symposium on
Conference_Location :
Fukuoka
Print_ISBN :
0-8186-8371-6
DOI :
10.1109/ISMVL.1998.679510
