エッジの向き, 運動方向, 速さを自己組織的に抽出する視覚系モデル

抄録

In this paper, we propose a neural network model that can detect in any of moving direction, moving speed and edge orientation of visual stimuli. The model adapts its structure to each function by a parameter of learning equations. The learning equations vary a horizontal connection and a self connection weight. Preferred stimulus is decided by the horizontal connection and the self connection. Computer simulations show self-organization of a edge orientation, a moving direction and a moving speed selectivity by single model. And we discuss a possibility of a functional specialization model.