Research map / Adversarial robustness and security / Attacks on text and graphs
Temporal Robustness against Data poisoning
Wenxiao Wang, Soheil Feizi
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 |
|---|---|---|---|
| increase training months n from 1 to 12 (ablation) | Base learner · News Category Dataset, all | clean accuracy: 54.09% -> 53.15% | Hurt |
| increase training months n from 1 to 12 (ablation) | Base learner · WORLD NEWS | clean accuracy: 59.00% -> 25.56% | Hurt |
| increase training months n from 1 to 12 (ablation) | Base learner · POLITICS | clean accuracy: 62.13% -> 53.96% | Hurt |
| increase training months n from 1 to 12 (ablation) | Base learner · TRAVEL | clean accuracy: 52.30% -> 62.26% | Improved |
| increase training months n from 1 to 12 (ablation) | Base learner · COLLEGE | clean accuracy: 42.30% -> 57.28% | Improved |
| increase base coverage n in temporal aggregation (ablation) | Temporal aggregation · News Category Dataset, all | clean accuracy | Hurt |
| increase base coverage n in temporal aggregation (ablation) | Temporal aggregation · News Category Dataset, bounded duration | temporal robustness | Hurt |
| increase aggregation size k from 6 to 24 (ablation) | Temporal aggregation · News Category Dataset, all | clean accuracy | Hurt |
| increase aggregation size k from 6 to 24 (ablation) | Temporal aggregation · News Category Dataset, bounded earliness and duration | temporal robustness | Improved |
| temporal aggregation with n=1, k=24 | Temporal aggregation · News Category Dataset | provable robustness (months of earliness / duration): 8~11 months earliness; 6~9 months duration | Improved |
Closest papers by meaning
- On Robustness of Linear Classifiers to Targeted Data Poisoning (AAAI 2026 · similarity 0.88)
- Accumulative Poisoning Attacks on Real-time Data (NeurIPS 2021 · similarity 0.87)
- Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples (NeurIPS 2026 · similarity 0.85)
- Scaling Trends for Data Poisoning in LLMs (AAAI 2025 · similarity 0.85)
- Provable Watermarking for Data Poisoning Attacks (NeurIPS 2025 · similarity 0.84)
- Phantom Transfer: Data Poisoning can Survive Data-Level Defences (NeurIPS 2026 · similarity 0.84)
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.
