Reproduce “APO2CS: Advantage-Aware Preference Optimization for Two-Level Constructive Search in 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
Kaixi Yang, Linbo Qiao, Wei Shang, Yanqi Shi, Dongsheng Li
APO2CS trains a constructive solver for combinatorial optimization by comparing complete solutions and short decision segments. It uses quality differences to scale preference signals at both levels. In a TSP ablation at 100 nodes, adding segment comparisons lowers the optimality gap from 0.36% to 0.10% versus whole-solution comparisons alone, supporting complementarity in this tested setting.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.