脳の機能局在をペトリネットでモデル化する方式の提案

抄録

This paper is a first step to develop a new control system which has better human like abilities than the conventional control system, and proposes a new method of modelling of brain's function distribution. By comparison with the neuro network which is structured homogeniously, proposed model is based on the petri net which is composed of state and transition and in the petri net, it is introduced learning and self-organizing capabilities which result from modified back-propagation and Hebb like method. But, it is the fundamental different point from the neuro network that learning and self-organizing of the proposed method are carried out on the only network pass of the token transfer. From simulation results, it has been cleared that proposed model can realize brain like function distribution, and therefore, in order to obtain strong non-linear functions, proposed model is superior to the neuro network.