Abstract—This paper considers a basic autonomous multi-agent model of group formation, whose agents carry feature vectors and meet each other at random in a free moving space. The agents combine to form groups, when their feature vectors are matched or their compatibilities are higher than a certain threshold. Similarly, if the compatibilities of the feature vectors between two groups are higher than the threshold not only between two agents but also between groups, the two groups are united into one. On the other hand, forming groups reduces a satisfaction of the groups by getting groups larger. This paper experimentally shows that, for a given threshold, there is an optimal threshold that maximizes the amount of satisfactions among groups, while the compatibility keeps the threshold, and the optimum threshold is related to the size of feature vectors, but the optimum threshold is independent to the number of agents.
Index Terms—Multi-agents, group formation, optimum threshold on satisfaction.
Kodai Shinkawa and Isamu Shioya are with the Faculty of Science and Engineering, Hosei University, Kajino-cho 3-7-2, Koganei-shi, Tokyo, Japan (e-mail: kodai.shinkawa.5z@stu.hosei.ac.jp; shioyai@hosei.ac.jp).
[PDF]
Cite:Kodai Shinkawa and Isamu Shioya, "Chaos Behavior in Group Formation," International Journal of Computer Theory and Engineering vol. 11, no. 2, pp. 27-30, 2019.
