Adapting in an uncertain world

This paper presents a general approach to creating an adaptive controller in uncertain conditions. Conventional adaptive processes rely on an underlying model (either implicit or explicit) to guide the adaptive process. In particular they rely on the deterministic nature of the process that is being controlled. The aim of this paper is to outline an adaptive process that can tackle such situations. The approach taken is to use a fuzzy logic controller, which can evolve (using a genetic algorithm) as it acquires knowledge of its domain. To illustrate this combination of fuzzy logic control [Smith F.S., 2001] with genetic algorithms, initial results on game playing are presented. The results show that the evolutionary fuzzy logic controller (EFLC), quickly learns aspects of strategy in card playing, even though it started with no strategies. The work presented in this paper could be extended to a wide range of domains, from contract bidding to guiding a diagnostic process.