Reproduce “Efficient and Optimal Algorithms for Contextual Dueling Bandits under Realizability” 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

Aadirupa Saha, Akshay Krishnamurthy

The paper presents an algorithm for contextual dueling bandits: learners choose two actions from a context and receive noisy pairwise preferences rather than rewards. It uses an online square-loss regression oracle to predict preferences and select randomized comparisons. Under realizability and oracle assumptions, it proves a square-root-T-type best-response regret bound with polynomial-in-K computational overhead, not an empirical result.

Summary by alphaXiv.

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