Research map / Bandits and online learning / Multi-armed bandits
Does Stochastic Gradient really succeed for bandits?
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's abstract only. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Studies whether the stochastic gradient bandit (SGB) policy with constant learning rate achieves logarithmic regret beyond the small-learning-rate regime, building on Mei et al. (2023, 2024).
Problem
Stochastic Gradient Bandit policies are known to converge asymptotically with a constant learning rate and to attain logarithmic regret only for sufficiently small learning rates; it is unclear whether logarithmic regret holds for larger learning rates.
Approach
Examines the SGB policy beyond small learning rates to determine whether logarithmic regret can hold for larger constant learning rates, extending the theory of Mei et al. (2023, 2024).
Experimental setup
Metrics: regret
Builds on: Mei et al. 2023 stochastic gradient bandit analysis Mei et al. 2024 stochastic gradient bandit analysis
Closest papers by meaning
- Small steps no more: Global convergence of stochastic gradient bandits for arbitrary learning rates (NeurIPS 2024 · similarity 0.92)
- p-Mean Regret for Stochastic Bandits (AAAI 2025 · similarity 0.90)
- Logistic Bandits with $\tilde{O}(\sqrt{dT})$ Regret without Context Diversity Assumptions (NeurIPS 2026 · similarity 0.90)
- Bayesian Regret Minimization in Offline Bandits (ICML 2024 · similarity 0.90)
- Beyond Primal-Dual Methods in Bandits with Stochastic and Adversarial Constraints (NeurIPS 2024 · similarity 0.90)
- Adaptive Regret for Bandits Made Possible: Two Queries Suffice (ICLR 2024 · similarity 0.90)
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.