Apr 27
Sharpness-Aware Poisoning: Enhancing Transferability of Injective Attacks on Recommender Systems
★★★★★
significance 2/5
Researchers propose a new method called Sharpness-Aware Poisoning (SharpAP) to improve the effectiveness of injective attacks on recommender systems. The method uses sharpness-aware minimization to identify and optimize against worst-case victim models, overcoming the limitations of traditional surrogate-based attacks.
Why it matters
Optimizing attacks against worst-case model architectures exposes a critical vulnerability in the robustness of next-generation recommendation engines.
Tags
#recommender systems #poisoning attacks #adversarial machine learning #transferabilityRelated coverage
- Global South OpportunitiesPivotal Research Fellowship 2026 (Q3): AI Safety Research Opportunity - Global South Opportunities
- arXiv cs.AIAn Intelligent Fault Diagnosis Method for General Aviation Aircraft Based on Multi-Fidelity Digital Twin and FMEA Knowledge Enhancement
- arXiv cs.AIPExA: Parallel Exploration Agent for Complex Text-to-SQL
- arXiv cs.AIThe Power of Power Law: Asymmetry Enables Compositional Reasoning
- arXiv cs.AIOn the Existence of an Inverse Solution for Preference-Based Reductions in Argumentation