Research map / Bandits and online learning / Online convex optimization and games
On Optimal Learning Under Targeted Data Poisoning
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 |
|---|---|---|---|
| Use SPV with the One-inclusion graph learner | Realizable learning under instance-targeted poisoning · Realizable distributions over H | η-adversarial risk: ≤ 42 η d | Helped |
| Allow instance-targeted poisoning of budget η | Any learner for a non-trivial class of VC dimension d · Realizable distributions over non-trivial H | η-adversarial risk: ≥ min{c₂ η d, 1/100} | Hurt |
| Use SPV with an agnostic One-inclusion graph learner | Agnostic learning under instance-targeted poisoning · Agnostic distributions over H | η-adversarial risk: ≤ 6C OPT + 42 C η d | Helped |
| Require a deterministic learner to achieve additive agnostic regret under targeted poisoning | Any H with at least two hypotheses · Agnostic distributions supported on two examples | η-adversarial risk: ≥ 2OPT+Ω(η')-O(1/n) | Hurt |
| Maintain properness by projecting the majority vote back into H using the projection-number technique | Realizable learning for halfspaces · Realizable distributions over halfspaces in ℝ^m | η-adversarial risk: ≤ 20 η d^3 | Helped |
Closest papers by meaning
- Agnostic Learning under Targeted Poisoning: Optimal Rates and the Role of Randomness (NeurIPS 2025 · similarity 0.94)
- What Distributions are Robust to Indiscriminate Poisoning Attacks for Linear Learners? (NeurIPS 2023 · similarity 0.89)
- Not All Poisons are Created Equal: Robust Training against Data Poisoning (ICML 2022 · similarity 0.88)
- Model-Targeted Poisoning Attacks with Provable Convergence (ICML 2021 · similarity 0.88)
- Private Online Learning against an Adaptive Adversary: Realizable and Agnostic Settings (NeurIPS 2025 · similarity 0.86)
- On the Learnability of Distribution Classes with Adaptive Adversaries (ICML 2025 · similarity 0.85)
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.