概要:

This study aimed to determine the proper amount of weight change ($\Delta w$) for dynamical recall.

Incremental Learning was applied for this pattern learning with the size from 100 to 350. And the effect on $\Delta w$ in dynamical recall was examined in each state.

The result, if the influence of the pattern to study was excluded, suggested the existence of the value of almost constant $\Delta w$ that succeeds in the most dynamical recall in the case of a lot of pattern size.

And there was a proportional relationship between the pattern size and the number of maximum dynamical recall successes, and the proportionality coefficient was 0.67.