Reproduce “ILP-BO: Integer Linear Programming-Based Black-Box 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
Hyakka Nakada, Shu Tanaka
ILP-BO uses exact integer linear programming to choose candidates in binary Bayesian optimization. It encodes finite-valued kernel distances with one-hot variables, turning nonlinear surrogate optimization into a 0–1 ILP without approximating the kernel. On Number Partitioning at D=30, ILP-BO Single reached the exact optimum in 0.6 of runs, versus 0.4 for BO-UCB. This certifies surrogate candidate selection, not the unknown objective.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.