拡張カルマンフィルタによる超高速ニューロ学習 : 第2報,文字パターン認識と文字パターン連想への適用(<小特集> インテリジェント・システム)

概要

論文の詳細を見る
In this paper, we discuss the application of the extended Kalman filter algorithm to the training of neural networks for pattern recognition and pattern formation. In neuron training, we use our acceleration technique, i. e. the VC technique, introduced in the first report. It is shown that the relationships between the dot pattern of the letter symbol and the binary code of the symbol are constructed effectively in both networks. Finally, it is also shown that pattern association from the incomplete pattern is realized by connecting the above two neural networks.
社団法人日本機械学会の論文
1993-08-25