Presentation 2021-10-28
A Study on Improvement Learning Performance with Chaos Neurons
Renshi Nagasawa, Masahiro Nakagawa,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In the backpropagation method in neural networks, the problem is that the energy converges to the local minimum. On the other hand, the chaotic neural networks, which is introduced chaotic dynamics found in the nervous system in vivo, was reported to have the ability to avoid local minimum. In this paper, we constructed a new model of the chaotic neural network by applying Chebyshev-type function as the activation function. We compared the learning ability of the constructed model with that of conventional periodic chaotic neuron models and so on. As a specific learning problem, we took up the n-bit parity problem. Each model was then applied to a three-layer neural network for training. As a result, the learning ability of the constructed model is not as good as that of the conventional periodic chaotic neuron model. The differences between the two results are discussed by showing the distribution of the output.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Chaos / Neural Networks / Back Propagation / n-bit Parity Problem
Paper # NC2021-23
Date of Issue 2021-10-21 (NC)

Conference Information
Committee MBE / NC
Conference Date 2021/10/28(2days)
Place (in Japanese) (See Japanese page)
Place (in English) Online
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Ryuhei Okuno(Setsunan Univ.) / Rieko Osu(Waseda Univ.)
Vice Chair Junichi Hori(Niigata Univ.) / Hiroshi Yamakawa(Univ of Tokyo)
Secretary Junichi Hori(Osaka Electro-Communication Univ) / Hiroshi Yamakawa(ATR)
Assistant Jun Akazawa(Meiji Univ. of Integrative Medicine) / Emi Yuda(Tohoku Univ) / Nobuhiko Wagatsuma(Toho Univ.) / Tomoki Kurikawa(KMU)

Paper Information
Registration To Technical Committee on ME and Bio Cybernetics / Technical Committee on Neurocomputing
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) A Study on Improvement Learning Performance with Chaos Neurons
Sub Title (in English)
Keyword(1) Chaos
Keyword(2) Neural Networks
Keyword(3) Back Propagation
Keyword(4) n-bit Parity Problem
1st Author's Name Renshi Nagasawa
1st Author's Affiliation Nagaoka University of Technology(NUT)
2nd Author's Name Masahiro Nakagawa
2nd Author's Affiliation Nagaoka University of Technology(NUT)
Date 2021-10-28
Paper # NC2021-23
Volume (vol) vol.121
Number (no) NC-223
Page pp.pp.28-33(NC),
#Pages 6
Date of Issue 2021-10-21 (NC)