Adaptive Neural Network Algorithm with Reinforcement Learning

Neural network algorithm (NNA) is a new type of population-based meta-heuristic algorithm, which is inspired by artificial neural networks. NNA is a result of the deep integration of artificial neural networks and metaheuristic algorithms. Benefiting from the unique structure of artificial neural networks, NNA shows excellent optimization performance. However, NNA has the disadvantages of slow convergence speed and easy trapping in local optimal solutions when solving complex multimodal optimization problems. To enhance the global search ability of NNA and expand the application of artificial neural networks in the field of metaheuristic algorithms, this paper proposes a new variant of NNA, namely adaptive neural network algorithm with reinforcement learning (ANNA). The core components of ANNA are the designed bias factor based on reinforcement learning and multiple learning transfer mechanism. To investigate the performance of ANNA, ANNA is employed to solve the challenging CEC 2014 test suite. Experimental results prove the validity of the improved strategies and excellent performance of ANNA in solving complex multi-modal optimization problems.