Research map / Bandits and online learning / Online convex optimization and games
Prediction with Corrupted Expert Advice
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 adaptive MW with decreasing step sizes η_t=√(log(N)/t), without knowing Δ or C | Prediction with expert advice · Adversarially-corrupted stochastic experts | expected pseudo regret: O(log(N)/Δ + C) | Helped |
| Use fixed-step MW with η=Δ/2 when the gap is known | Prediction with expert advice · Adversarially-corrupted stochastic experts | expected pseudo regret: 4 log N/Δ + 4C | Helped |
| Replace adaptive MW's FTRL update with its OMD update (ablation) | Adaptive MW · Stochastic experts (no corruption) | expected pseudo regret: O(Δ^{-1} log(N) log^2(Δ^{-1} log N)) | Mixed |
| Replace adaptive MW's FTRL update with its OMD update (ablation) | Adaptive MW · Adversarially-corrupted stochastic experts | expected pseudo regret: Ω(Δ T₁), T₁ = Θ(min{C/Δ², exp(√C/α)}) | Hurt |
| Add adversarial corruption to the feedback | Adaptive MW (OMD) · Adversarially-corrupted stochastic experts | expected pseudo regret: Ω(C/Δ) | Hurt |
| Use adaptive MW (FTRL) rather than OMD when feedback is corrupted (ablation) | Adaptive MW · Adversarially-corrupted stochastic experts | expected pseudo regret: O(log(N)/Δ + C); OMD: Ω(C/Δ) | Helped |
| Use adaptive MW (OMD) rather than FTRL without corruption (ablation) | Adaptive MW · Stochastic experts (no corruption) | pseudo regret | Mixed |
Closest papers by meaning
- Prediction with expert advice under additive noise (NeurIPS 2025 · similarity 0.89)
- When Lower-Order Terms Dominate: Adaptive Expert Algorithms for Heavy-Tailed Losses (NeurIPS 2025 · similarity 0.87)
- Bandits with Abstention under Expert Advice (NeurIPS 2024 · similarity 0.87)
- Online Mixture of Experts: No-Regret Learning for Optimal Collective Decision-Making (NeurIPS 2025 · similarity 0.86)
- Principled Bayesian Optimization in Collaboration with Human Experts (NeurIPS 2024 · similarity 0.86)
- Learning from Imperfect Human Feedback: A Tale from Corruption-Robust Dueling (ICLR 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.