STUDY ON NATURAL FILTER SELECTION STRATEGY IMPROVEMENT METHOD IN SEAMO2 TO SOLVE OPTIMIZATION MAXIMUM OBJECTIVES PROBLEMS

  • Tran Hai Thanh*, Le Hoang Hiep, Doan Ngoc Phuong
Keywords: Computer science; The problem of multi-purpose bag; Individual; Random; Population

Abstract

The paper focuses on studying, analyzing and evaluating some cases occurring in the strategy of selecting and replacing individuals in the population of the SEAMO2 algorithm to find solutions to improve the algorithm to solve the multi-objective bag problem. By using the substitution of the random selection in the algorithm's substitution strategy is the worst (most substitution) instance in a limited space of instances. In comparison with the old method performed before, the new improvement of SEAMO2_LG shows that in the first generations of the population (or when running the algorithm with short runs), then finding suitable individuals to replace will help the population converge to the Pareto boundary faster or when running with long runs (large number of generations), in the last generations (when the population has reached the optimal threshold), the restriction of search space will save time.

điểm /   đánh giá
Published
2022-08-26
Section
NATURAL SCIENCE – ENGINEERING – TECHNOLOGY