Improving the Learner Classifier System with a Basic Memetic Algorithm for Rule-Based Problem Solving
نویسندگان
1 MSc student, Medical Biotechnology Research Center, Ashkezar Branch, Islamic Azad University of Ashkezar, Yazd, Iran, Email: phdmrdma@gmail.com
2 Ph.D. Student in Computer Engineering Software, Faculty of Computer Engineering, Islamic Azad University of Maybod, Maybod, Iran, Email: m.r.dehghani.m.a@gmail.com
doi
چکیده
Memetic algorithms are used to optimize the expensive target performance. The evaluation of the current population number is conducted by searching in the previous generations and preserving the values by the memetic algorithm. A significant number of generations are required to find the optimal value of the objective function in rule-based systems. The learning classifier system is one of the methods of generating value and classification for law. Each rule includes a set of properties. The function of the learning classifier systems is based on the genetic algorithm that it is not possible to search and save the previous steps in order to find a better solution to the problem. In this article, the memetic algorithm is used to improve and optimize the learner classifier system. In the proposed system, the memetic algorithm is used to create a population to improve the learning classifier system in the state space. The efficiency, convergence speed, and standard deviation of the proposed method are revealed using the implementation. The results indicated that the proposed hybrid method of replacing the memetic algorithm in the learning classifier system can significantly speed up the system and improve the quality to maintain better rules according to the search of previous generations.
