Research map / Adversarial robustness and security / Attacks on text and graphs
Untargeted Attack against Federated Recommendation Systems via Poisonous Item Embeddings and the Defense
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 ClusterAttack instead of no attack | MF FedRec · ML-1M | HR@5: 0.03549 -> 0.02451 | Hurt |
| apply ClusterAttack instead of Fang | MF FedRec · ML-1M | HR@5: 0.03038 -> 0.02451 | Hurt |
| apply ClusterAttack instead of no attack | SASRec FedRec · Gowalla | NDCG@5: 0.02217 -> 0.01544 | Hurt |
| add contrastive learning task to benign-client training | FedRec with UNION · ML-1M | HR@5 | Helped |
| add contrastive learning to ClusterAttack (ablation) | MF + MultiKrum+UNION · ML-1M | HR@5: 0.03378 -> 0.03525 | Helped |
| add contrastive learning to ClusterAttack (ablation) | MF + NormBound+UNION · ML-1M | HR@5: 0.03449 -> 0.03566 | Helped |
| remove adaptive clustering (ablation) | ClusterAttack · ML-1M | HR@5 | Hurt |
| use LabelFlip instead of no attack | MF FedRec · ML-1M | HR@5: 0.03549 -> 0.03561 | No clear effect |
| use Gaussian instead of no attack | MF FedRec · ML-1M | HR@5: 0.03549 -> 0.03555 | No clear effect |
| increase malicious-client ratio from 0.5% to 1% | ClusterAttack · ML-1M | HR@5: 0.03001 -> 0.02451 | Hurt |
| combine MultiKrum with UNION | FedRec under ClusterAttack · ML-1M | HR@5 | Helped |
| combine NormBound with UNION | FedRec under ClusterAttack · ML-1M | HR@5 | Helped |
Closest papers by meaning
- Rethinking Byzantine Robustness in Federated Recommendation from Sparse Aggregation Perspective (AAAI 2025 · similarity 0.87)
- SecEmb: Sparsity-Aware Secure Federated Learning of On-Device Recommender System with Large Embedding (ICML 2025 · similarity 0.87)
- Federated Recommendation with Additive Personalization (ICLR 2024 · similarity 0.84)
- SPMC: Self-Purifying Federated Backdoor Defense via Margin Contribution (ICML 2025 · similarity 0.83)
- Parameter Disparities Dissection for Backdoor Defense in Heterogeneous Federated Learning (NeurIPS 2024 · similarity 0.83)
- Tracing Back the Malicious Clients in Poisoning Attacks to Federated Learning (NeurIPS 2025 · 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.