Research map / Bandits and online learning / Online convex optimization and games
Learning Strategy-Aware Linear Classifiers
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 Grinder's adaptive polytope partitioning and in-oracle | Strategic classification against δ-bounded myopically rational agents · Simulated discrete and continuous action spaces | Regret or cumulative loss | Helped |
| Use Grinder rather than EXP3 | Strategic classification against δ-bounded myopically rational agents · Simulated discrete action space | Regret | Helped |
| Use an accurate in-oracle rather than a regression-based approximate oracle (ablation) | Grinder · Simulated discrete action space | Regret | Helped |
| Increase the agents' manipulation radius δ | Grinder · Simulated strategic classification | Regret or cumulative loss | Hurt |
| Use Grinder rather than BGD | Strategic classification against δ-bounded myopically rational agents · Simulation | Regret or cumulative loss | Helped |
| Use a convex surrogate such as hinge loss despite strategic best responses | Linear classifiers against δ-bounded myopically rational agents · Hinge-loss counterexample | Hinge loss: 0.95 > 0.5·0.99 + 0.5·0.875 | Mixed |
| Use a sequence with sublinear Stackelberg regret | Learner in the constructed two-action instance · Strategic classification | External regret: 0.15T | Hurt |
| Use a sequence with sublinear external regret | Learner in the constructed two-action instance · Strategic classification | Stackelberg regret: 0.05T | Hurt |
| Allow an ε-approximate in-oracle with ε ≤ 1/√T rather than requiring an accurate oracle | Grinder · Strategic classification, Stackelberg regret | Regret guarantee: \varepsilon \leq 1/\sqrt{T} | No clear effect |
Closest papers by meaning
- Strategic Classification With Externalities (ICLR 2025 · similarity 0.89)
- Impact of Decentralized Learning on Player Utilities in Stackelberg Games (ICML 2024 · similarity 0.87)
- One-Shot Strategic Classification Under Unknown Costs (ICML 2024 · similarity 0.87)
- Strategic Linear Contextual Bandits (NeurIPS 2024 · similarity 0.87)
- Bayesian Strategic Classification (NeurIPS 2024 · similarity 0.87)
- Strategic Littlestone Dimension: Improved Bounds on Online Strategic Classification (NeurIPS 2024 · similarity 0.87)
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.