General information

This repository contains the implementation of paper Subset Selection for Evolutionary Multi-Objective Optimization. It includes our GSEMO-ACC framework. Part of codes use the open-source toolkit from HisaoLabSUSTC in https://github.com/HisaoLabSUSTC/BenchSS.

Code

Algorithms for Quality Indicators

Code for GSEMO-HV: GSEMOHV.m
function [RecordHV,RecordR2,Subset,time]=GSEMOHV(PopObj,selNum,r,W)
+ PopObj: the set of candidate solutions.
+ selNum: the constraint on the number of selected solutions.
+ r: reference point.
+ W: the set of direction vectors for hypervolume approximate indicator.


GSEMO-ACC-IGD: GSEMOIGD.m
function [Record,Subset,time]=GSEMOIGD(PopObj,selNum)
+ PopObj: the set of candidate solutions.
+ selNum: the constraint on the number of selected solutions.

Code for GSEMO-ACC-IGD+: GSEMOIGDp.m
function [Record,Subset,time]=GSEMOIGDp(PopObj,selNum)
+ PopObj: the set of candidate solutions.
+ selNum: the constraint on the number of selected solutions.

Code for GSEMO-ACC-R2: GSEMOR2Tchebycheff.m
function [RecordR2Tch,Subset,time]=GSEMOR2Tchebycheff(PopObj,selNum,r,W)
PopObj: the set of candidate solutions.
selNum: the constraint on the number of selected solutions.
r: Utopian point.
W: the set of weight vectors for R2 indicator.

Corresponding Greedy Algorithms for Quality Indicators

GreedyHVSelection.m
LazyIGDSelection.m
LazyIGDpSelection.m
LazyR2TchebycheffSelection.m

Dataset

Dataset files for section V-C: Results on Benchmark and Real-world Problems are in subfolder './datasets/Experiments on Benchmark and Real-world Problems/'
+ Pareto front dataset files in Benchmark test suite are in subfolder '/PF/'
+ Dataset files generated by MOEAs in Benchmark test suite are in subfolder '/EMOA/'
+ Real-world problem dataset files are in subfolder '/REAL-WORLD PROBLEM DATASETS/'

Dataset files for section V-D: Scalability are in subfolder './datasets/Experiments on the Scalability/'

Dataset files for section V-F: What If Considering More Advanced MOEAs? are in subfolder './datasets/Experiments of NSGAII and MOEAD/'

Citation

If you find our work helpful, please cite:

@ARTICLE{10079175,
  author={Gu, Yu-Ran and Bian, Chao and Li, Miqing and Qian, Chao},
  journal={IEEE Transactions on Evolutionary Computation}, 
  title={Subset Selection for Evolutionary Multi-Objective Optimization}, 
  year={2023},
  volume={},
  number={},
  pages={1-1},
  keywords={Approximation algorithms;Optimization;Statistics;Sociology;Greedy algorithms;Linear programming;Evolutionary computation;Subset selection;Evolutionary multi-objective optimization;Quality indicators;Submodularity;multi-objective evolutionary optimizer (MOEAs);global simple evolutionary multi-objective algorithm (GSEMO)},
  doi={10.1109/TEVC.2023.3261134}}