A Learned Multi-objective Bacterial Foraging Optimization Algorithm with Continuous Deep Q-Learning.

Multi-objective bacterial foraging optimization (MBFO) algorithm is a kind of efficient swarm intelligence algorithm for multi-objective problems. However, both nested loop and the swimming step length setting are crucial to capability improvement for the specific problem. To deal with these two problems, this paper proposes learned multi-objective bacterial foraging optimization algorithm (LMBFO) which transforms nested loop between operations into a conditional selection and firstly combined multi-objective bacterial foraging optimization algorithm with continuous deep q-learning to realize adaptive parameter control. To verify the feasibility of LMBFO, it is trained and tested on classical multi-objective benchmark functions. Moreover, MBFO is utilized as the comparison to illustrate the superiority of our proposed MLBFO.