Evolving classification of agents’ behaviors: a general approach

Abstract

By recognizing the behavior of others, many different tasks can be performed, such as to predict their future behavior, to coordinate with them or to assist them. If this behavior recognition can be done automatically, it can be very useful in many applications. However, an agents’ behavior is not necessarily fixed but rather it evolves/changes. Thus, it is essential to take into account these changes in any behavior recognition system. In this paper, we present a general approach to the classification of streaming data which represent a specific agent behavior based on evolving systems. The experiment results show that an evolving system based on our approach can efficiently model and recognize different behaviors in very different domains, in particular, UNIX command-line data streams, and intelligent home environments.