Research map / Bandits and online learning / Multi-armed bandits
Boltzmann Exploration Done Right
Nicolò Cesa-Bianchi, Claudio Gentile, Gabor Lugosi, Gergely Neu
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Start with a constant learning rate, then use a gap- and horizon-dependent increasing schedule | Boltzmann exploration · Bounded-reward stochastic multi-armed bandits | regret: R_T ≤ 16eK log T/Δ² + 9K/Δ² | Improved |
| Replace empirical means with influence-function reward estimates | BGE · Heavy-tailed stochastic multi-armed bandits with bounded second moments | regret: The bound coincides with that of Theorem [ref], except that σ² is replaced by V. | Improved |
| Use standard Boltzmann variants rather than uncertainty-scaled perturbations | Boltzmann exploration · Bernoulli bandit with malicious initialization | identification of the optimal arm | Hurt |
| Use BGE instead of standard Boltzmann exploration | Boltzmann exploration · Bernoulli bandit with malicious initialization | empirical performance | Improved |
Closest papers by meaning
- MaxInfoRL: Boosting exploration in reinforcement learning through information gain maximization (ICLR 2025 · similarity 0.90)
- Provable and Practical: Efficient Exploration in Reinforcement Learning via Langevin Monte Carlo (ICLR 2024 · similarity 0.89)
- Bayesian Exploration Networks (ICML 2024 · similarity 0.88)
- Forced Exploration in Bandit Problems (AAAI 2024 · similarity 0.87)
- Feel-Good Thompson Sampling for Contextual Bandits: a Markov Chain Monte Carlo Showdown (NeurIPS 2025 · similarity 0.87)
- Small steps no more: Global convergence of stochastic gradient bandits for arbitrary learning rates (NeurIPS 2024 · similarity 0.87)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
