ファジィペトリネットによる好感度変化学習パラメタのHopfield型ファジィニューラルネットワークを用いた適応的探索の提案

摘要

Emotion Generating Calculations (EGC) method can measure pleasure/displeasure from Favorite Value (FV) of words in sentences. FV is the value of the taste information. However, FV is changed according to time and situation. Therefore, the learning method of FV is required. Ichimura et al. proposed the adaptive learning method according to the user’s current emotion or mood. The method was the reasoning of Fuzzy Petri Net with some predefined parameter. In this paper, we propose the automaticaly learning method of parameter by using Hopfield Neural Network. We report some experimented results for 7 subjects.