Multi User Learning Agent on the Distribution of MDPs

Recently, the researches of social robots and agents that smoothly communicate to human are paid to attention. The conventional researches seem to have aimed at the achievement of the society in the meaning of the wide sense. In a word, it is an approach where the designer embedded social competence like human. In this paper, we propose model-based multi user learning agent (M-MULA) to interact with various types of people effectively towards realization of social interaction. M-MULA is equipped with two functions for social learning. One is direct learning function by individualizing of user parameters, and another is indirect learning function based on past experience and the similarity between each users. We confirm proposed method in computer simulation to evaluate the effectiveness