Summary
International Symposium on Nonlinear Theory and its Applications
2009
Session Number:A3L-D
Session:
Number:A3L-D1
The introduction of the brain-inspired systems
Kiyohisa Natsume, Tetsuo Furukawa,
pp.-
Publication Date:2009/10/18
Online ISSN:2188-5079
Summary:
David Marr described three levels of information processing in the brain, that is, the computational level, the algorithmic level, and the implementation level. Many brain science and engineering researchers are ordering their study based on these levels. Note that the last level is not, for example, the process of making chips or devices but the study of how the neuronal networks in a brain actually implement an algorithm in the brain. The study of the brain-inspired systems (BrainIS) is based on Marr’s three levels with the addition of implementation of the brain algorithm on chips or devices and the checking of whether the algorithm actually works in the real world. In the study of BrainIS, you draw the algorithm from the results of the brain science or hypothesize the putative brain algorithm, implement the algorithm on chips or devices, and check whether the brain algorithm works well in the real world using the platform rather than simply mimicking the brain’s information processing. When you draw the algorithm from the results of neuroscience, you can make an inspired algorithm from the results. In BrainIS, you not only describe a simple brain function but also a whole inspired system consisting of brain functions for the information processing that is part of those functions. The inspired system autonomously moves, learns the environment, and behaves using an environment-system loop.