Neural network is the technology which reproduced human brain artificially. Time-sereies such as sound are widely studied using this. Neural network with internal memory is for a learning of time-sereies.
In this research, Neural network with internal memory is made to learn complicated time‐sereies, and is inputted various signals to investigate about condition of internal memory.
As the result,this network outputs right signal with phase differences, after it outputs different signal from supervisory signal. But, a supervisory signal is reproducible correctly. Because, internal memory layer have connection weights from the middle layer, so internal memory have variable output every time. It help to recall. We show the condition of internal memory transition in a two‐dimensional display. As a result the condition of internal memory transition is relevant to a supervisory signal. And the condition of internal memory behaves as the time-sereies. The condition of internal memory transition is formed by distinction of a supervisory signal.