We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

econ.EM

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Economics > Econometrics

Title: Risk and optimal policies in bandit experiments

Abstract: We provide a decision theoretic analysis of bandit experiments. Working within the framework of diffusion asymptotics, we define suitable notions of asymptotic Bayes and minimax risk for these experiments. For normally distributed rewards, the minimal Bayes risk can be characterized as the solution to a second-order partial differential equation (PDE). Using a limit of experiments approach, we show that this PDE characterization also holds asymptotically under both parametric and non-parametric distributions of the rewards. The approach further describes the state variables it is asymptotically sufficient to restrict attention to, and thereby suggests a practical strategy for dimension reduction. The PDEs characterizing minimal Bayes risk can be solved efficiently using sparse matrix routines. We derive the optimal Bayes and minimax policies from their numerical solutions. These optimal policies substantially dominate existing methods such as Thompson sampling and UCB, often by a factor of two. The framework also covers time discounting and pure exploration.
Subjects: Econometrics (econ.EM); Machine Learning (cs.LG)
Cite as: arXiv:2112.06363 [econ.EM]
  (or arXiv:2112.06363v12 [econ.EM] for this version)

Submission history

From: Karun Adusumilli [view email]
[v1] Mon, 13 Dec 2021 00:41:19 GMT (374kb,D)
[v2] Tue, 12 Apr 2022 05:50:27 GMT (1132kb,D)
[v3] Mon, 18 Apr 2022 06:03:32 GMT (1132kb,D)
[v4] Mon, 2 May 2022 16:36:40 GMT (1168kb,D)
[v5] Sat, 14 May 2022 22:32:10 GMT (1149kb,D)
[v6] Tue, 14 Jun 2022 10:05:07 GMT (1149kb,D)
[v7] Fri, 8 Jul 2022 20:49:59 GMT (1149kb,D)
[v8] Thu, 4 Aug 2022 14:45:53 GMT (1149kb,D)
[v9] Tue, 9 Aug 2022 09:08:44 GMT (1149kb,D)
[v10] Mon, 15 Aug 2022 20:51:20 GMT (1150kb,D)
[v11] Thu, 18 Aug 2022 01:51:12 GMT (1150kb,D)
[v12] Fri, 16 Sep 2022 23:24:16 GMT (1150kb,D)

Link back to: arXiv, form interface, contact.