Neurocomputers, non Neuman type computers, contrived at the same time as the Neuman type computers which is the main stream at present, declined temporarily. However, it is paid attention as the software works on Neuman type computer with rapid growth of Neuman type computer. Because, with simulating neurocomputer, Neuman type computer can process informations like a brain.
Neurocomputers can recognize letters, pictures and so on, and estimate of the chaotic informations like temperature, stock prices and so on. This is easy for a brain and not for Neuman type computers. However, at present neurocomputers are simulated by Neuman type computers. Therefore, yet, it is impossible for neurocomputers to process informations as well as brains. But it is interesting that Neuman type computers process information flexibly like neurocomputers.
The chaos is remarkable as the way of making computer process information flexibility like a human being.
In this study, a chaotic neural network learn time series signals. The chaotic neural network is the extension of the neural network and resembles a nervous cell in the brain more than the neural network. It can process the time series signals. Therefore, experiment is carried out as follows. A chaotic neural network is made to input music and learn it. And then it is made to output the learned music or the similar one.
This experiment use the back propagation method. And for a chaotic neural network to learn efficiency, learning constant is examined.