This paper describes an artificial immune algorithm (IA) combined with estimation of distribution algorithm (EDA), named IA-EDA, for the traveling salesman problem (TSP). Two components are incorporated in IA-EDA to further improve the performance of the conventional IA. First, aiming to strengthen the information exchange during different solutions, two kinds of EDAs involving univariate marginal distribution algorithm and population-based incremental learning are altered based on the permutation representation of TSP. It is expected that new promising candidate solutions can be sampled from the constructed probabilistic model of EDA. Second, a heuristic refinement local search operator is proposed to repair the infeasible solutions sampled by EDA. Therefore, IA-EDA can alleviate the deficiencies of the conventional IA and can find better solutions for TSP by well balancing the exploitation and exploration of the search. Experiments are conducted based on a number of benchmark instances with size up to 100 000 cities. Simulation results show that IA-EDA is effective for improving the performance of the conventional IA and can produce better or competitive solutions than other hybrid algorithms. (Non-member) received the M.S. degree from the University of Toyama (UT), Toyama, Japan, in 2011. He is currently working toward the Ph.D. degree in the Faculty of Engineering, UT. His research interests include artificial immune algorithms, evolutionary algorithms, and combinatorial optimizations. Yirui Wang (Non-member) received the B.S. degree from Donghua University (DU), Shanghai, China, in 2014. He is currently working toward the M.S. degree at the College of Information Sciences and Technology, DU. His current research interests include computational intelligence, swarm intelligent algorithms, and combinatorial optimizations.
Zhe Xu
S153IEEJ Trans 11(S1): S142-S154 (2016) Z. XU ET AL.Sheng Li (Non-member) received the B.S. degree from Nanjing