A new bidding strategy in LCS using a decentralized loaning and bid history

In a strength based learning classifier systems (LCS), auctioning among classifiers that match to an environmental message has been used as a way of identifying winner classifiers. All classifiers participating in an auction issue a bid proportional to their strength and a winner classifier is allowed to fire and receive a reward or punishment from its environment as a consequence of its action. In this kind of bidding strategy, good classifiers with low strength and little experience have to wait until the strength of less useful classifiers has come down through continuous taxation. This slows down the convergence of the learning system to the optimal solution sets. In addition, offspring classifiers that come from weak parents as a result of randomness in the selection process may inherit a small strength as compared to experienced classifiers in the population. A mutation occurring at a point may however make them better match to more environmental inputs. But due to a low initial strength they have to wait for some time till they mature and try their action. This paper introduced a decentralized loaning approach to mitigate the above shortcomings of the bidding strategy in traditional LCS. Loaning among classifiers in the population is allowed. In direct analogy with real auctions, all classifiers matching the current input compare the average bid history with their potential bid based on their current strength. The average bid history parameter gives general information about the bid market (potential of competent classifiers) and determines the amount of loan a classifier should ask. The results obtained show a significant improvement on the performance of the system.