| 研究実績の概要 |
As an intermediate level approach between AI and neuroscience, behaviors of artificial neural network have been elucidated. However, since their functions have not been enough elucidated, we are studying the function of the neural network focused on signal transmission or communication function.
In 2D mesh neural network simulation, the signal is propagated in the form of spike waves with fluctuations. The corresponding receiving neuron group is able to identify the signal after having learned to form an asynchronous multiplex communication channel such as 9:9. Grouping and synchronic firing is often seen in natural neuronal networks and seems to be effective for stable/robust communication in conjunction with spatial multiplex communication.
Corresponding to these, we showed we can set communication channels such as 3:n in experiments on cultured neuronal networks. There, stimulated neurons could be classified/recognized at several receiving neurons away from the stimulated neurons as to what kind of stimulation was added based on the arrived spike waves. Through these studies, we also found that utilizing correlations between some adjacent neurons (grouping) improves the quality of the classification for communication, similar to a diversity antenna, which is used to improve the quality of communication in artificial data communication systems, and similarity to sound identification by the ear.
Both of the simulation in artificial neural networks and the experiment in cultured neuronal networks support the multiplex communication scheme in the brain.
|
