Pittsburgh-style learning classifier system for multiple environments: towards robust waterbus route for several situations
by Keiji Sato; Keiki Takadama
International Journal of Bio-Inspired Computation (IJBIC), Vol. 3, No. 6, 2011

Abstract: This paper proposes an accuracy-based Pittsburgh-style learning classifier system (LCS) that can find effective and robust solutions against several different situations, and aims at investigating its effectiveness in the waterbus route optimisation problem. For this purpose, our accuracy-based Pittsburgh-style LCS: 1) introduces a new fitness calculation to remain robust classifiers (i.e., solutions) in different situations 2) employs NSGA-II to find the most effective and robust solutions among a lot of Pareto front solutions found in the multi-objective optimisation. Through intensive simulations on the waterbus route optimisation problem, we have revealed that our proposed LCS can find the waterbus routes that can cope with two different situations. In detail: 1) the relative fitness calculation can find the robust routes in comparison with the ordinary fitness calculation 2) the accuracy-based selection of the parents succeeds to find more effective and robust route in the different environments in comparison with the NSGA-II-based selection in the multi-objective optimisation.

Online publication date: Mon, 07-Nov-2011

The full text of this article is only available to individual subscribers or to users at subscribing institutions.

 
Existing subscribers:
Go to Inderscience Online Journals to access the Full Text of this article.

Pay per view:
If you are not a subscriber and you just want to read the full contents of this article, buy online access here.

Complimentary Subscribers, Editors or Members of the Editorial Board of the International Journal of Bio-Inspired Computation (IJBIC):
Login with your Inderscience username and password:

    Username:        Password:         

Forgotten your password?


Want to subscribe?
A subscription gives you complete access to all articles in the current issue, as well as to all articles in the previous three years (where applicable). See our Orders page to subscribe.

If you still need assistance, please email subs@inderscience.com