Research map / Bandits and online learning / Multi-armed bandits
Adversarial Attacks on Adversarial Bandits
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 |
|---|---|---|---|
| Apply the easy template-based attack targeting a1 | Exp3 · Easy attack example, T=10^6 | target arm selections: 1.45× 10^4 -> 9.85× 10^5 | Helped |
| Apply the easy template-based attack | ExpRb with Φ=T^0.5 · Easy attack example, T=10^6 | target arm selections: 9.83× 10^6 | Helped |
| Use the general attack instead of the easy attack | Exp3 · Easy attack example, T=10^6 | target arm selections: 9.85× 10^5 -> 9.85× 10^5 | No clear effect |
| Increase ε from 0.25 to 0.4 (ablation) | Exp3 general attack · General attack example | target arm selections | Helped |
| Set ε=0.25 (ablation) | Exp3 general attack · General attack example | cumulative attack cost: 4.20× 10^2, 2.84× 10^3, 1.85× 10^4, and 1.14× 10^5 | Helped |
| Increase ExpRb's assumed attack budget Φ (ablation) | ExpRb under the easy attack · Easy attack example, T=10^6 | target arm selections: 98.3%, 89.7%, and 63.2% | Hurt |
| Increase ExpRb's assumed attack budget Φ (ablation) | ExpRb under the general attack · General attack example, T=10^6 | cumulative attack cost: 1.40×10^5, 3.62× 10^5, and 5.15× 10^5 | Hurt |
| Use the general attack with ε=0.25 | Exp3 · General attack example, T=10^6 | target arm selections: 9.24× 10^5 | Helped |
Closest papers by meaning
- Adversarial Attacks on Stochastic Bandits (NeurIPS 2018 · similarity 0.93)
- Stochastic Bandits Robust to Adversarial Attacks (ICLR 2025 · similarity 0.91)
- Adversarial Bandits with Corruptions: Regret Lower Bound and No-regret Algorithm (NeurIPS 2020 · similarity 0.91)
- Practical Adversarial Attacks on Stochastic Bandits via Fake Data Injection (NeurIPS 2026 · similarity 0.90)
- Faster Rates for Private Adversarial Bandits (ICML 2025 · similarity 0.89)
- Taming Heavy-Tailed Losses in Adversarial Bandits and the Best-of-Both-Worlds Setting (NeurIPS 2024 · similarity 0.88)
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.