矢内 浩文（PY） 玉川大学 工学部
Hiro-F. Yanai, Faculty of Engineering, Tamagawa University
E-mail: yanai@eng.tamagawa.ac.jp
There are two phases in the dynamics of neural nets: neuronal and synaptic. Through the extension of linear version of 2-stage neuron, I show that neuronal dynamics can approximate the orthogonal learning (synaptic dynamics) to any desired precision. Although the two phases are equivalent there also are interesting differences.
Keywords: ダイナミクス，シナプス可塑性，相関行列，直交射影行列