COOPERATIVE-BASED LEARNING CLASSIFIER SYSTEM:A MAN-MACHINE MODEL APPROACH

Abstract

In this paper, we propose a Cooperative-Based Learning Classifier System (CBLCS) which combines the genetic-based learning classifier system with a Man-Machine Model (MMM) to come up with better performance than traditional classifier system. The proposed MMM consists of two constiuents: Conflict Resolution Mechanism and User Interface Component. In particular, we use a Reliability Probability to decide whether to take the user's advice or not when conflicts occur. An elevator-scheduling problem is then used to demonstrate the performance of the proposed CBLCS approach. Results show that the CBLCS can improve not only the learning speed but also the solution quality.