Reproduce “Evolving Neural Solvers for Multi-objective Combinatorial Optimization” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Ya-Hui Jia, Le-Xin Wen, Yang Wang, Wei-Neng Chen, Yi Mei

ENS addresses multi-objective combinatorial optimization by expanding a small set of preference-specific anchor solvers into a population of variants. It evolves only the decoder’s query projection and retains candidates contributing non-dominated solutions, so retained solvers can run without an inference-time preference. Across 12 tested settings, ENS has the best hypervolume in 10; on Tri-TSP100, ENS-Aug scores 0.5122 versus WS-LKH’s 0.5076.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.