ランダム神経ネットワークにおける秩序の形成 : 脳機能の解明に向けて

概要

論文の詳細を見る
The behavior of a simple neural network model is discussed. We suppose that each neuron is coupled to a certain number of randomly chosen neurons. Each neuron is regarded as a binary decision element, and it uodates its states ("firing" or "resting") synchronously in discrete time step according to a deterministic rule. It is found that this network exhibits strong order, i.e., a dominant mode of excitation always appears regardless of initial conditions. The ratio of neurons responsible for such dominant mode is at most about 40 present. This fact suggests that the neural mechanism of self-organization (i.e., a reconstruction of neural network where large number of neurons are "deleted" by apoptosis) occurred in the early stage of embryology of the brain.
青森公立大学の論文
2005-03-22