Research map / Adversarial robustness and security / Attacks on text and graphs
Potent but Stealthy: Rethink Profile Pollution Against Sequential Recommendation via Bi-Level Constrained Reinforcement Paradigm
Jiajie Su, Zihan Nan, Yunshan Ma, Xiaobo Xia, XiaoHua Feng, Weiming Liu, Xiang Chen, Xiaolin Zheng, Chaochao Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (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 |
|---|---|---|---|
| add directionality reward (ablation) | CREAT · ML-1M (NARM) | H@10: 0.0117 -> 0.0157 | Improved |
| add diversity reward (ablation) | CREAT · ML-1M (NARM) | H@10: 0.0143 -> 0.0157 | Improved |
| add pattern inversion reward (dir and div) (ablation) | CREAT · ML-1M (NARM) | H@10: 0.0025 -> 0.0157 | Improved |
| add distribution consistency reward (ablation) | CREAT · ML-1M (NARM) | H@10: 0.0271 -> 0.0157 | Hurt |
| add distribution consistency reward (ablation) | CREAT · Beauty (NARM) | H@10: 0.6291 -> 0.5633 | Hurt |
| add distribution consistency reward (ablation) | CREAT · Beauty (BERT4Rec) | H@10: 0.7461 -> 0.6885 | Hurt |
| add pattern inversion reward (ablation) | CREAT · Beauty (BERT4Rec) | H@10: 0.5044 -> 0.6885 | Improved |
| add directionality reward (ablation) | CREAT · Beauty (BERT4Rec) | N@10: 0.4812 -> 0.6285 | Improved |
| add diversity reward (ablation) | CREAT · Beauty (BERT4Rec) | H@10: 0.6603 -> 0.6885 | Improved |
| use dynamic barrier instead of fixed penalty 0.1 to 2.0 (ablation) | CREAT · attack performance | attack performance | Improved |
| use GRPO instead of REINFORCE (ablation) | CREAT · reward trajectories | pattern inversion reward | Improved |
| use CREAT instead of best baseline | attack · Beauty (BERT4Rec) | HR@10: 0.0109 -> 0.2601 | Improved |
Closest papers by meaning
- Coordinated Attacks against Contextual Bandits: Fundamental Limits and Defense Mechanisms (ICML 2022 · similarity 0.85)
- On the Unexpected Effectiveness of Reinforcement Learning for Sequential Recommendation (ICML 2024 · similarity 0.84)
- Sim4Rec: Data-Free Model Extraction Attack on Sequential Recommendation (AAAI 2025 · similarity 0.84)
- Recommender Ecosystems: A Mechanism Design Perspective on Holistic Modeling and Optimization (AAAI 2024 · similarity 0.84)
- UNEX-RL: Reinforcing Long-Term Rewards in Multi-Stage Recommender Systems with UNidirectional EXecution (AAAI 2024 · similarity 0.83)
- Strategic Linear Contextual Bandits (NeurIPS 2024 · similarity 0.83)
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.
