In recent years, neural network is used in many fields, such as image processing and pattern recognition.
In this research, search access is studied using a chaotic neural network. Search access is the reference method using the feature of patterns. Incremental learning is used to study search access, especially on dynamical recall.
As a result of research, the dynamical recall, the major focus of this study, was not realized, while a cycle was observed in the recall patterns. Its recall patterns cycle appeared in the same order as each pattern was learned. The cycle still appeared using a random order.
Secondly, presynaptic inhibition was added on various conditions, but no good periodic recall was obtained.
Lastly, the periodic recall was shifted to the self-recall status in which a specific pattern is kept recalling. And that self-recall status was successfully observed.