A phenomenon called dynamical recall is observed on a chaotic neural network composed by chaotic neurons which show chaotic reactions. A chaotic neural network which completed learning of patterns irregularly recalls learned patterns without external inputs, and this is a phenomenon called dynamical recall. An increase of patterns recalled by dynamical recall is very important to realize a technology called search access.
Experiments to increase patterns recalled by dynamical recall were conducted in this study. In the first experiment, I examined whether an increase of the number of neurons brings an increase of the number of recalled patterns or not. The results of the first experiment showed that an increase of the number of neurons brings an increase of the number of recalled patterns, while the results showed that an excessive increase of the number of neurons brings a decrease of the number of recalled patterns. It was expected that a decrease of a fixed number about learning (amount of weight change) brings an increase of the number of recalled patterns on the chaotic neural network composed by a large number of neurons, the second experiment was conducted to judge whether this expectation is right or wrong. The results of the second experiment showed that this expectation is right on condition that amount of weight change is moderately big.