A High-Performance Self-Learning Antelopes Migration Algorithm (SAMA) for Global Optimization

Abstract

In global optimization, Nature-Inspired Computing (NIC) has attracted increasing attention in recent years. However, a common predicament for most optimization algorithms is the poor tradeoff between global exploration and local exploitation, which results in premature convergence. Moreover, the complex mechanism of algorithms makes them inconvenient to use for actual users, who are not specialists in computational intelligence (CI). To handle these defects, this paper proposes a succinct high-performance nature-inspired algorithm for global optimization, namely, Self-learning Antelopes Migration Algorithm (SAMA), which imitates the migration behavior of Tibetan antelopes on the HohXil grassland of China. The antelope herd always migrates to the most fertile place they have found, where ordinary members graze about here and some vigorous members selected as scouts to explore farther to seek a better place. The interesting behaviors of different antelopes guarantee the development and best state of the herd. They also have learning capability which adaptively corrects their searching strategy through the past migration experience. To verify the high performance of the proposed SAMA, we compare our proposed algorithm with five advanced nature-inspired algorithms. A series of benchmark tests demonstrate that our SAMA has remarkable superiority on both optimization accuracy and speed.