• Mathematics, Computer Science
  • Published in J. Mach. Learn. Res. 2014

An Information-Theoretic Analysis of Thompson Sampling

@article{Russo2014AnIA,
  title={An Information-Theoretic Analysis of Thompson Sampling},
  author={Daniel Russo and Benjamin Van Roy},
  journal={J. Mach. Learn. Res.},
  year={2014},
  volume={17},
  pages={68:1-68:30}
}
We provide an information-theoretic analysis of Thompson sampling that applies across a broad range of online optimization problems in which a decision-maker must learn from partial feedback. This analysis inherits the simplicity and elegance of information theory and leads to regret bounds that scale with the entropy of the optimal-action distribution. This strengthens preexisting results and yields new insight into how information improves performance. 

Citations

Publications citing this paper.
SHOWING 1-10 OF 97 CITATIONS

Simple Algorithms for Dueling Bandits

VIEW 7 EXCERPTS
CITES METHODS & BACKGROUND
HIGHLY INFLUENCED

An Information-Theoretic Analysis for Thompson Sampling with Many Actions

VIEW 10 EXCERPTS
CITES BACKGROUND & METHODS

An Information-Theoretic Analysis of Thompson Sampling for Large Action Spaces

VIEW 9 EXCERPTS
CITES BACKGROUND & METHODS

Myopic Bayesian Design of Experiments via Posterior Sampling and Probabilistic Programming

VIEW 6 EXCERPTS
CITES BACKGROUND, RESULTS & METHODS
HIGHLY INFLUENCED

Satisficing in Time-Sensitive Bandit Learning

VIEW 12 EXCERPTS
CITES METHODS & BACKGROUND

Thompson Sampling for Bandit Problems

VIEW 5 EXCERPTS
CITES RESULTS & BACKGROUND
HIGHLY INFLUENCED

Learning to Optimize via Information-Directed Sampling

VIEW 8 EXCERPTS
CITES BACKGROUND & METHODS

Thompson Sampling with Approximate Inference

VIEW 3 EXCERPTS
CITES BACKGROUND
HIGHLY INFLUENCED

Online Experimentation for Information Retrieval

VIEW 4 EXCERPTS
CITES BACKGROUND & METHODS
HIGHLY INFLUENCED

FILTER CITATIONS BY YEAR

2012
2020

CITATION STATISTICS

  • 5 Highly Influenced Citations

  • Averaged 25 Citations per year from 2017 through 2019

References

Publications referenced by this paper.
SHOWING 1-10 OF 30 REFERENCES

Learning to Optimize via Posterior Sampling

VIEW 6 EXCERPTS

Regret Analysis of Stochastic and Nonstochastic Multi-armed Bandit Problems

VIEW 3 EXCERPTS
HIGHLY INFLUENTIAL

Entropy and Information Theory

VIEW 5 EXCERPTS
HIGHLY INFLUENTIAL

Prior-free and prior-dependent regret bounds for Thompson Sampling

VIEW 3 EXCERPTS
HIGHLY INFLUENTIAL

Further Optimal Regret Bounds for Thompson Sampling

VIEW 4 EXCERPTS
HIGHLY INFLUENTIAL

Analysis of Thompson Sampling for the Multi-armed Bandit Problem

VIEW 4 EXCERPTS
HIGHLY INFLUENTIAL

Overview of content experiments: Multi-armed bandit experiments, 2014

  • S. L. Scott
  • URL https://support.google.com/analytics/answer/2844870?hl=en. [Online; accessed 9-February-
  • 2014
VIEW 1 EXCERPT

Generalized Thompson Sampling for Contextual Bandits

  • Lihong Li
  • Mathematics, Computer Science
  • ArXiv
  • 2013
VIEW 1 EXCERPT