A deep learning guided memetic framework for graph coloring problems

作者：

Highlights：

• We propose a deep learning guided memetic framework for graph coloring problems.

• The approach is applied to solve two popular coloring problems — vertex coloring and weighted coloring.

• The approach shows excellent results on benchmark graphs, including 14 improved best results.

• Experiments are conducted to assess the benefits of deep learning for the proposed approach.

摘要

•We propose a deep learning guided memetic framework for graph coloring problems.•The approach is applied to solve two popular coloring problems — vertex coloring and weighted coloring.•The approach shows excellent results on benchmark graphs, including 14 improved best results.•Experiments are conducted to assess the benefits of deep learning for the proposed approach.

论文关键词：Population-based search,GPU-based parallel search,Deep learning,Heuristics,Graph coloring

论文评审过程：Received 19 October 2021, Revised 3 October 2022, Accepted 4 October 2022, Available online 10 October 2022, Version of Record 22 October 2022.

论文官网地址：https://doi.org/10.1016/j.knosys.2022.109986