非線形振動子場における局在振動の研究(機械学習によるバイオデータマインニング,生命現象の非線形性,一般)  [in Japanese] Localized Oscillatory Excitation on Non-Linear Oscillatory Fields by Radial Isochron Clocks  [in Japanese]

不安定な平衡解を囲む安定な周期解をもつ非線形振動子の中で最も簡単なものとして知られるRIC(Radial Isochron Clock)またはSL(Stuart-Landau)振動子を用いて,非線形振動子場のモデルを作成した.シミュレーションにより,二つのメキシカンハット型結合をもつ振動子場が,外部入力の複数の極大付近に局在振動を安定に保持し,各局在振動内では同位相に,そして各局在振動間では逆位相に引き込むことがわかった.このモデルは,結び付け問題を発生させることなく情報分離を行う振動子型自己組織モデルの基礎を与える.

We propose a model of the non-linear oscillatory field, using Radial Isochron Clocks (RICs) or Stuart-Landau (SL) oscillators, which is the simplest dynamical system that has one stable limit cycle around one unstable equilibrium. Our computer simulation showed that the non-linear oscillatory field with two kinds of Mexican-hat-type connection could keep two or more than two localized oscillatory excitation areas stably around maximal points of an external inputs, and could also realize in-phase phase-locking within a localized oscillatory excitation area, but anti-phase phase-locking between different localized oscillatory excitation areas. This non-linear oscillatory field provides a base for an oscillatory self-organizing neural model which separates and extracts several kinds of information without binding problelm.