Budgeted Bayesian Bandits for Future Utilization

Key points

• In this paper, we study the variant where exploration has a hard budget and precedes exploitation.1
• Prior guarantees include a 1/3 approximation by Guha and Munagala using Lagrangian duality, and a 4/27 approximation by Ma against a primal linear programming relaxation for a more general model.1
• We show that the analyses of both algorithms are tight, even when their relaxations are asymptotically exact.1
• We then provide an efficient algorithm whose guarantee approaches 1-e^{-1} approx 0.63 for bounded exchangeable reward observations in a non-vanishing regime, which is asymptotically tight against natural linear programming relaxations.1

Sources (1)

• [1]Budgeted Bayesian Bandits for Future Utilization
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 16, 07:29 PM

  The multi-armed bandit problem models the trade-off between exploration and exploitation.

  In this paper, we study the variant where exploration has a hard budget and precedes exploitation.

Extractive summary: sentences quoted from the sources.