Computer Simulation of Fractal Transition on Recurrent Network

• SATO Shozo

  Faculty of Engineering, Hokkaido University

• GOHARA Kazutoshi

  Faculty of Engineering, Hokkaido University

Bibliographic Information

Other Title

• リカレントネットワークにおけるフラクタル遷移の実験的検証

Abstract

We have to understand behaviors of Recurrent Neural Netwaorks(RNN) considerd non-automous systems when external inputs change with time. In this case, we had shown that a behavior of RNN are considered a sequence of transitions between attractors corresponding external input patterns. This paper experimentally shows that a trajectory of this sequence in the state space produce a fractal structure, and presents the RNN is a dynamical system on a fractal-like invariant set.

Journal

• IEICE technical report. Nonlinear problems

  IEICE technical report. Nonlinear problems 96 (509), 123-130, 1997-02-06

  The Institute of Electronics, Information and Communication Engineers

Keywords

• Recurrent Network
• Dynamical System
• Fractal Transition

Details

• CRID: 1570572702385691776
• NII Article ID: 110003291669
• NII Book ID: AN10060800
• Text Lang: ja
• Data Source:
  • CiNii Articles