Research map / Bandits and online learning / Online convex optimization and games
Strategic Apple Tasting
Keegan Harris, Chara Podimata, Steven Wu
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 |
|---|---|---|---|
| Shift the linear policy to account for strategic behavior and learn only from clean data | greedy linear policy · strategic classification with apple tasting feedback, stochastic contexts | strategic regret bound | Improved |
| Learn from clean data only under one-sided feedback | greedy algorithm · strategic classification with apple tasting feedback | strategic regret bound | Hurt |
| Replace the greedy algorithm with strategy-aware explore-then-commit | greedy algorithm · strategic classification with apple tasting feedback, small or unknown T | strategic regret bound: d * T^{2/3} strategic regret | Improved |
| Combine greedy and explore-then-commit with a doubling trick | both algorithms · strategic classification with apple tasting feedback, unknown T | strategic regret bound: min{c(d, delta) sqrt(T), d T^{2/3}} | Improved |
| Run EXP3 over a discretized grid of strategy-aware policies | adversarial agent contexts · strategic classification with apple tasting feedback, adversarial contexts | strategic regret bound: T^{(d+1)/(d+2)} strategic regret | Improved |
| Adversarial instead of stochastic contexts | apple tasting algorithms · strategic classification with apple tasting feedback | strategic regret bound: T^{(d+1)/(d+2)} vs sqrt(T) | Hurt |
| Overestimate the manipulation parameter delta | greedy algorithm · strategic classification with apple tasting feedback | strategic regret bound | Hurt |
| Extend one-sided feedback to bandit feedback with unknown r(0) | all algorithms · strategic classification with apple tasting feedback | strategic regret bound | No clear effect |
Closest papers by meaning
- Strategic Linear Contextual Bandits (NeurIPS 2024 · similarity 0.88)
- Strategyproof Reinforcement Learning from Human Feedback (NeurIPS 2025 · similarity 0.87)
- Incentivized Learning in Principal-Agent Bandit Games (ICML 2024 · similarity 0.86)
- Bandits Meet Mechanism Design to Combat Clickbait in Online Recommendation (ICLR 2024 · similarity 0.86)
- No-Regret Learning Under Adversarial Resource Constraints: A Spending Plan Is All You Need! (NeurIPS 2025 · similarity 0.86)
- Bandits with Ranking Feedback (NeurIPS 2024 · similarity 0.86)
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.
