An Improved Teaching-Learning-Based Optimization for Multitask Optimization Problems

In recent years, multitask optimization (MTO) plays an important role in real life, covering various fields such as engineering, finance, and agriculture. MTO can solve multiple optimization problems in the meantime and heighten the performance of solving each task. The teaching-learning-based optimization (TLBO) algorithm and its improved variants focus on solving single problem. In this article, a novel multitask teaching-learning-based optimization (MTTLBO) algorithm is proposed to handle multitask optimization (MTO) problems. Firstly, MTTLBO makes full use of knowledge transfer between different optimization problems to improve performance and efficiency on solving MTO problems. The concept of opposition-based learning (OBL) is introduced into the teaching stage which allow students to receive a wide range of knowledge and find individuals with great differences. MTTLBO is compared with some advanced evolutionary multitask algorithms. The experimental results state clearly that the MTTLBO algorithm owns excellent performance on nine sets of single-objective problems.