Youthful Teacher Jing Wang from the School of Software and Internet of Things Engineering Publishes Significant Research Outcomes in Top SCI Journal ELSEVIER

​​

  Recently,JingWang,a young teacher from the School of Software and Internet of ThingsEngineering, published a research paper titled"Anovel multi-state reinforcement learning-based multi-objective evolutionary algorithm"in the internationally renowned journal ELSEVIER. The paper introduces a multi-objective evolutionary algorithm (MRL-MOEA) basedon multi-state reinforcement learning, which combines the decision-making capabilities of reinforcement learning with the demands of multi-objective optimization, significantly enhancing the performance of the algorithm.

　The main work of the research includes three aspects: First, by constructing a state model based on the distribution of individualsin the objective space and using a reinforcement learning frame work to dynamically select the optimal crossover operator, a balance between diversity and convergence is achieved; Second, in response tothe insufficient convergence in certain areas of the Pareto front, a strategy for adjusting weight vectors has been proposed to improve the distribution in sparse areas and achieve a more uniform Paretofront; Finally, the performance of MRL-MOEA was widely verified onmultiple benchmark test sets including WFG and DTLZ, proving the algorithm's excellent solving ability and competitive advantage when dealing with problems involving 3 to 10 objectives. The experiment alresults fully prove the practicality and superiority of the algorithmin complex multi-objective optimization.

　Teacher JingWang's research achievement has not only attracted widespread attention inthe academic community but also provided new research ideas andmethods for the field of multi-objective optimization, which has important the oretical and practical application value.