Research map / Bandits and online learning / Online convex optimization and games
An Algorithm for Stochastic and Adversarial Bandits with Switching Costs
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Group rounds into increasing-length blocks and run Tsallis-INF once per block | Tsallis-Switch · Oblivious adversarial bandits with fixed switching cost | pseudo-regret with switching costs | Helped |
| Use Tsallis-Switch rather than Tsallis-INF without blocks | Stochastic losses, easy setting (Δ = 0.2, λ = 0.025) · Stochastic losses, easy setting | pseudo-regret | Hurt |
| Use Tsallis-Switch rather than Tsallis-INF without blocks | Stochastic losses, zero switching cost (λ = 0, Δ = 0.05) · Stochastic losses, zero switching cost | pseudo-regret | No clear effect |
| Use Tsallis-Switch rather than EXP3 or BaSE | Stochastic losses, zero switching cost (λ = 0, Δ = 0.05) · Stochastic losses, zero switching cost | pseudo-regret | Helped |
| Use Tsallis-Switch rather than BaSE | Stochastically constrained adversarial losses · Stochastically constrained adversarial regime | pseudo-regret | Helped |
| Use BaSE rather than Tsallis-Switch | Adversarial losses · Adversarial regime | pseudo-regret | Hurt |
| Set switching cost to zero | Tsallis-Switch · Stochastically constrained adversarial regime | pseudo-regret | Helped |
Closest papers by meaning
- Better Best of Both Worlds Bounds for Bandits with Switching Costs (NeurIPS 2022 · similarity 0.94)
- Stochastic Bandits Robust to Adversarial Attacks (ICLR 2025 · similarity 0.91)
- Hybrid Regret Bounds for Combinatorial Semi-Bandits and Adversarial Linear Bandits (NeurIPS 2021 · similarity 0.91)
- Sparsity-Agnostic Linear Bandits with Adaptive Adversaries (NeurIPS 2024 · similarity 0.90)
- Beyond Primal-Dual Methods in Bandits with Stochastic and Adversarial Constraints (NeurIPS 2024 · similarity 0.89)
- Improved Best-of-Both-Worlds Regret for Bandits with Delayed Feedback (NeurIPS 2025 · similarity 0.89)
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.