Information Processing through Dynamics――Recent Advances in Reservoir Computing

Bibliographic Information

Other Title
  • ダイナミクスによる情報処理――レザバー計算の最近の発展
  • 人工知能と物理学 ダイナミクスによる情報処理 : レザバー計算の最近の発展
  • ジンコウ チノウ ト ブツリガク ダイナミクス ニ ヨル ジョウホウ ショリ : レザバー ケイサン ノ サイキン ノ ハッテン

Search this article

Abstract

<p>Reservoir computing is a framework for computation such that:(1) an input signal is given to a dynamical system (reservoir); (2) the input sequence is encoded by the high-dimensional dynamics of the reservoir; and (3) the desired output is formed by a simple readout mechanism of the reservoir. Recently, physical reservoir computing, which utilizes real physical systems as the reservoir, has attracted attention in the fields of nonlinear physics and machine learning. As an application of reservoir computing, we report the characteristics of the reservoir composed of neural networks which are optimized by recurrent informax principle.</p>

Journal

  • Butsuri

    Butsuri 74 (5), 306-313, 2019-05-05

    The Physical Society of Japan

Related Projects

See more

Details 詳細情報について

Report a problem

Back to top