A METHOD OF PROCESSING DATA FLOWS IN A NEURAL NETWORK, AND A NEURAL NETWORK

摘要

There are neural network external sensor (2) to indicate to be environmentally isolated with to receive parameter. Parameter indicates weighted and fed and be input to host node (1), whose output, which transmits the parameter, indicates other weighted type input external effect device (5). The output of outer executor is fed back to the influence after the external sensor of input terminal to environment. In addition, parameter indicates feedback directly from sensor host node, the feedback that then sum of parameters indicates is formed. This causes quick and extremely complex neural network to be learnt, and is learning, and carry out task is carried out without the use of feedback signal by environment. Therefore, the present invention a kind of neural network similitude is provided can be where close to the human consciousness.