Negatively Correlated Search for Constrained Optimization

Yuan Li; Xiaofen Lu; Xin Yao

10.1109/CEC53210.2023.10254141

2023

2023 IEEE Congress on Evolutionary Computation

979-8-3503-1459-5

2023 IEEE Congress on Evolutionary Computation (CEC)

1-10

1-5 July 2023

Chicago, IL, USA

USA

Evolutionary algorithms (EAs) combined with constraint handling techniques (CHTs) are very effective in solving constrained optimization problems (COPs). Recently, negatively correlated search (NCS) has been shown to be powerful for real-world multimodal optimization problems (MMOP). However, there has been few works on utilizing NCS to solve COPs. In this paper, we present a novel constrained optimization evolutionary algorithm (COEA) named NCS-E which combined NCS with a new CHT ensemble method to deal with COPs. It integrates three complementary CHTs as voters, and individuals are first voted by each voter, then the individual with more weighted votes is considered better. In addition, the negative correlation information of the voted individuals is utilized to adaptively adjust the voting weights. To demonstrate the performance of NCS-E, NCS-E is compared with four NCS-based COEAs and two state-of-the-art methods, ∊MAgES and MAgES-VMCH, on 57 real-world COPs. The statistical results show that NCS-E exhibits the best performance among the compared methods. Moreover, NCS-E was applied to solve the Deep Neural Network pruning problem and compared with the state-of-the-art pruning method OLMP on three reference models. Empirical results show that the proposed method improves the performance of OLMP on all three DNN pruning tasks.

Negatively Correlated Search Constrained Optimization Evolutionary Algorithm Ensemble DNN Pruning

第一 ; 通讯

英语

EI

20234314944546

Constraint Handling ; Deep Neural Networks ; Evolutionary Algorithms

Ergonomics And Human Factors Engineering:461.4 ; Computer Theory, Includes Formal Logic, Automata Theory, Switching Theory, Programming Theory:721.1 ; Systems Science:961

IEEE

Department of Computer Science and Engineering, The Guangdong Provincial Key Laboratory of Brain-Inspired Intelligent Computation, Southern University of Science and Technology, Shenzhen, China

Yuan Li,Xiaofen Lu,Xin Yao. Negatively Correlated Search for Constrained Optimization[C],2023:1-10.