Modified teaching-learning based optimization for 0–1 knapsack optimization problems

In this article, a modified teaching-learning-based optimization (MTLBO) algorithm is proposed for solving 0-1 knapsack optimization problems. The MTLBO incorporated estimation of distribution operation into teaching phases, which aims at reducing the possibility of premature and predicting an elite teacher. A stochastic local exploitation is used in teaching phase for improving local searching capability. Moreover, in the learning phase, a new global learning operation is presented to boost learning efficiency. Several classic 0-1 knapsack cases are selected to evaluate the performance of MTLBO. Numerical results reveal that the proposed algorithm surpasses TLBO and several other promising heuristic methods.