View in EDS

Multitasking multiobjective optimization based on transfer component analysis.

Saved in:
Bibliographic Details
Title: Multitasking multiobjective optimization based on transfer component analysis.
Authors: Hu, Ziyu1,2 (AUTHOR) hzy@ysu.edu.cn, Li, Yulin1,2 (AUTHOR), Sun, Hao1,2 (AUTHOR), Ma, Xuemin1,2 (AUTHOR)
Source: Information Sciences. Aug2022, Vol. 605, p182-201. 20p.
Subject Terms: DIFFERENTIAL evolution, EVOLUTIONARY algorithms, BENCHMARK problems (Computer science), KNOWLEDGE transfer
Abstract: • In order to efficiently promote the positive transfer of knowledge between tasks, an explicit knowledge transfer strategy is designed. • To narrow the differences between individual distributions, a dimension reduction method based on TCA is used. • The potential relationship between tasks is analyzed based on individual characteristics. Multitasking optimization (MTO) has emerged as a new research topic in recent years. The purpose of MTO is to use the correlations between tasks to find a set of optimal solutions to simultaneously optimize multiple tasks. MTO research focuses on promoting positive transfer of knowledge and sufficient information exchange between tasks. To positively promote the efficiency of knowledge transfer, a multiobjective multifactorial evolutionary algorithm based on transfer component analysis (TCA) and differential evolution (DE) called TCADE is proposed. The TCA method is used to construct a dimensionality reduction subspace, in which the correlation between two tasks is used to find a set of solutions. Co-evolution of multiple populations is promoted after explicit transfer of the solutions. Furthermore, a DE operator is used to generate more diverse individuals. TCADE effectively utilizes the potential relationships between tasks to transfer solutions across them and promotes knowledge transfer between them. TCADE is tested by experiments on nine benchmark problems. The experimental results show that the proposed algorithm obtains 15 inverted generational distance optimal values for 18 test functions. [ABSTRACT FROM AUTHOR]
Copyright of Information Sciences is the property of Elsevier B.V. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Business Source Index
Description
Abstract:• In order to efficiently promote the positive transfer of knowledge between tasks, an explicit knowledge transfer strategy is designed. • To narrow the differences between individual distributions, a dimension reduction method based on TCA is used. • The potential relationship between tasks is analyzed based on individual characteristics. Multitasking optimization (MTO) has emerged as a new research topic in recent years. The purpose of MTO is to use the correlations between tasks to find a set of optimal solutions to simultaneously optimize multiple tasks. MTO research focuses on promoting positive transfer of knowledge and sufficient information exchange between tasks. To positively promote the efficiency of knowledge transfer, a multiobjective multifactorial evolutionary algorithm based on transfer component analysis (TCA) and differential evolution (DE) called TCADE is proposed. The TCA method is used to construct a dimensionality reduction subspace, in which the correlation between two tasks is used to find a set of solutions. Co-evolution of multiple populations is promoted after explicit transfer of the solutions. Furthermore, a DE operator is used to generate more diverse individuals. TCADE effectively utilizes the potential relationships between tasks to transfer solutions across them and promotes knowledge transfer between them. TCADE is tested by experiments on nine benchmark problems. The experimental results show that the proposed algorithm obtains 15 inverted generational distance optimal values for 18 test functions. [ABSTRACT FROM AUTHOR]
ISSN:00200255
DOI:10.1016/j.ins.2022.05.037