An Improved Most Valuable Player Algorithm with Twice Training Mechanism

The most valuable player algorithm is inspired from these players who want to win the Most Valuable Player (MVP) trophy, it have higher overall success percentage. Teaching-learning-based optimization (TLBO) simulates the process of teaching and learning. TLBO has fewer parameters that must be determined during the renewal process. This paper proposes twice training mechanism to enhance the search ability of the most valuable player algorithm (MVPA) through hybrid TLBO algorithm, and named it teaching the most valuable player algorithm (TMVPA). In TMVPA, designs two behaviors of training and abstract two training modes: pre-competition training and post-competition training. Before individual competition, join the pre-competition training to coordinated exploitation ability and the exploration ability of the original algorithm and join the post-competition training to prevent from falling into the local optimal field after the corporate competition. We test three benchmark functions and an engineering design problem. Results show that TMVPA has effectively raised algorithm accuracy.