時系列クラスタリングによる行動ルールの獲得と行動支援

抄録

In this paper, we propose a method which supports human-action-decision should be done in one’s circumstance. First of our proposed method, human’s action is modeled and acquired as Extended If-Then-Rules which is possible to treat Time-Series-Information. The rules are used to compare with a circumstance, therefore a best-matched rule is selected by RIFCDP and the rule is used to estimate the future circumstance and next action at each time.