One of the ways for learning to make Chaotic Neural Network do associative memory is Successive Learning. The past research of this laboratory prove that Successive Learning is better than the former Correlational Learning and that it is suited for learning. I investigated how many patterns it could learn using the Successive Learning.
Patterns to make Chaotic Neural Network learn used capital and small letters of the alphabet as patterns for learning in the past researches. I used patterns where white and black points were programed to exist at random to lessen inclination and resemblance for learning. I changed a condition of patterns to make it learn, and the number of learning, the frequency to repeat and learn as the way to make it learn. Then,I investigated what kind of special character would be seen in the number of learning patterns. By putting this special character to practical use, we can find a condition where efficient learning is made possible.
競合作用