An improved artificial tree algorithm with two populations (IATTP). (September 2021)

• Record Type: Journal Article
• Authors: Xiao, Yaping; Chi, Hanbin; Li, Qiqi
• Is Part Of: Engineering applications of artificial intelligence. Volume 104(2021)
• Publication Date: 2021-09
• Subjects: Artificial tree algorithm -- Two populations -- Competition mechanism -- Update operators; Engineering -- Data processing -- Periodicals; Artificial intelligence -- Periodicals; Expert systems (Computer science) -- Periodicals; Ingénierie -- Informatique -- Périodiques; Intelligence artificielle -- Périodiques; Systèmes experts (Informatique) -- Périodiques; Artificial intelligence; Engineering -- Data processing; Expert systems (Computer science); Periodicals; 620.00285
• DOI: 10.1016/j.engappai.2021.104324
• Languages: English
• ISSNs: 0952-1976

Abstract

Abstract: Many new bio-inspired algorithms are recently being proposed, artificial tree (AT) algorithm, inspired by the growth of trees and the update behavior of branches, is one of them. There are also some improved AT algorithms being proposed to improve their calculation accuracy. However, the main challenges of AT algorithms lie in the insufficiencies in the design of update operators as well as the position interaction between branches and the capture of key information and the performance of AT algorithms needs to be enhanced. This work proposes an improved AT algorithm with two populations (IATTP). In IATTP, the update strategies of branches are redesigned, and a variety of efficient update operators are designed and applied. The branch population is changed from one to two, and the competition mechanism between populations is proposed. Through the migration of branches between populations, the scale of population with better efficiency is expanded and the size of population with lower efficiency is reduced, thus a reasonable interaction between populations and branches is realized. With above strategies, the efficiency and accuracy of IATTP are significantly improved. The results of IATTP are proved to be advantageous when the performance of IATTP is compared with AT algorithm, improved artificial tree (IAT) algorithm and feedback artificial tree (FAT) algorithm through typical test problems. Meanwhile, the results of IATTP in current state are also preferable when IATTP is compared with other improved algorithms in high dimensional problems. The experimental results prove that IATTP is competitive in solving optimization problems. Highlights: An improved artificial tree algorithm named (IATTP) is proposed. The update operators of branches are redesigned. The competition mechanism of populations is proposed. The performance of IATTP is better than that of AT, IAT and FAT. IATTP performs better than some typical improved algorithms.