Reinforcement learning algorithm with network extension for pulse neural network

Author

Takita, Koichiro ; Osana, Yuko ; Hagiwara, Masafumi

Author_Institution

Fac. of Sci. & Technol., Keio Univ., Yokohama, Japan

Volume

4

fYear

2000

fDate

2000

Firstpage

2586

Abstract

In this paper, we propose a new hierarchical pulse neural network and its reinforcement learning algorithm with network extension. The proposed pulse neural network has three layers, and all of the neurons are pulse neurons. This network learns relations between input pulse sequences and the desired outputs by updating connection weights and by adding neurons dynamically. We carried out a computer simulation to confirm the performance of the proposed algorithm

Keywords

learning (artificial intelligence); neural nets; pulse circuits; sequences; virtual machines; algorithm performance; computer simulation; connection weight updating; dynamic neuron addition; hierarchical pulse neural network; input pulse sequences; input-output relation learning; network extension; pulse neurons; reinforcement learning algorithm; Assembly; Biological information theory; Biological neural networks; Biological system modeling; Computer architecture; Computer simulation; Information processing; Learning; Neural networks; Neurons;

fLanguage

English

Publisher

ieee

Conference_Titel

Systems, Man, and Cybernetics, 2000 IEEE International Conference on

Conference_Location

Nashville, TN

ISSN

1062-922X

Print_ISBN

0-7803-6583-6

Type

conf

DOI

10.1109/ICSMC.2000.884383

Filename

884383