Reproduce “Fast Iterative Combinatorial Auctions via Deep Reinforcement Learning” 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
Luis F. Paul Tosta, Quinn Arnold, Gianluca Brero
FastCombo learns offline price policies for iterative combinatorial auctions, choosing each price vector in one neural-network pass. In a controlled one-item, three-profile experiment, five policies averaged 1.65 rounds, versus 1.85 for a Bayesian rule. On CATS, price-selection computation was hundreds of times lower, while clearing was generally somewhat lower; these results do not establish general round savings or lower total auction time.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.