phenomenon · first seen 7 Oct 2026
adversarial attacks on bandits
Manipulating reward feedback or history so a bandit algorithm is misled into selecting an attacker-chosen target arm.
Mentions per week
PapersCodeDiscussion
Earliest sources
- Beyond Reward Suppression: Near-Optimal Offline Attacks on Warm-Start Bandits with Bounded RewardsPapers · 7 Oct 2026 · Qirun Zeng, Manhin Poon, Xiangxiang Dai, Qixin Zhang, Jinhang Zuo
Latest sources
- Beyond Reward Suppression: Near-Optimal Offline Attacks on Warm-Start Bandits with Bounded RewardsPapers · 7 Oct 2026 · Qirun Zeng, Manhin Poon, Xiangxiang Dai, Qixin Zhang, Jinhang Zuo