概要:

Chaotic neural network is a neural network which consists of chaotic neurons. Chaotic neuron is an artificial neuron model that includes chaotic behavior. Artificial neuron model is a computing system that imitates our brain.

The purpose of this study is to examine the success rate of search access using random patterns by chaotic neural network with incremental learning. Search access is a method of searching for target pattern using only features of pattern to search from the stored patterns. It is necessary for the search access to use presynaptic inhibition and feature extraction

In this study, randomness or trigonometric functions in presynaptic inhibition and expansion of feature extraction are introduced to get high success rate of search access with random patterns.

As a result, it was found that success rate of search access increased. Particularly in the case of expansion of feature extraction, the success rate of search access has increased significantly. The problem is that average time to converge in network becomes too large. To shorten the convergence time in network is the future task.