A new experiential learning electromagnetism-like mechanism for numerical optimization

The Electromagnetism-like Mechanism algorithm (EM) is a population-based search algorithm which has shown good achievements in solving various types of complex numerical optimization problems so far. To date, the study on experience-based local search mechanism is relatively limited, and there is no study in the literature to integrate experience-based features into the EM. This work introduces an experience-learning feature into the EM for the first time. A new Experiential Learning Electromagnetism-like Mechanism algorithm (ELEM) is proposed in this paper. The ELEM is integrated with two new components. The first component is the particle memory concept which allows the particles to remember the details of their past search experience. The second component is the experience analysing and decision making mechanisms which enables the particles to adjust the settings for the coming iterations. Combining the advantages of this strong exploitation strategy and the powerful exploration mechanism of the EM, the proposed ELEM strikes a good balance in providing well diversified solutions with high accuracy. The results from extensive numerical experiments carried out using 21 challenging test functions show that ELEM is able to provide very competitive solutions and significantly outperforms other optimization techniques. It can thus be concluded from the results that the proposed ELEM performs well in solving high dimensional numerical optimization problems.