Title :
A modified leaky integrator network for temporal pattern processing
Author :
Uchiyama, Tadasu ; Shimohara, Katsunori ; Tokunaga, Yukio
Author_Institution :
NTT Human Interface Lab., Kanagawa, Japan
Abstract :
A modified leaky integrator, that is, a neuron model that generates continuous output according to its activation history, is proposed for temporal pattern processing. A modified leaky integrator network is described, and its applicability to temporal pattern processing is discussed theoretically. The learning algorithm for the modified leaky integrator network is presented, and the efficiency in temporal pattern processing is shown through simulations.<>
Keywords :
learning systems; neural nets; pattern recognition; activation history; learning systems; modified leaky integrator network; neural nets; neuron model; pattern recognition; temporal pattern processing; Learning systems; Neural networks; Pattern recognition;
Conference_Titel :
Neural Networks, 1989. IJCNN., International Joint Conference on
Conference_Location :
Washington, DC, USA
DOI :
10.1109/IJCNN.1989.118621
