Research map / Multi-agent systems and games / Games and equilibria
Beyond Rational Illusion: Behaviorally Realistic Strategic Classification
Xinpeng Lv, Yunxin Mao, Renzhe Xu, Chunyuan Zheng, Yikai Chen, Haoxuan Li, Yang Shi, Jinxuan Yang, Yuanlong Chen, Yuanxing Zhang, Shaowu Yang, Wenjing Yang and 2 more
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 |
|---|---|---|---|
| add prospect-guided modeling (Pro-SF) vs rational classifier | Pro-SF · Adult (non-rational agents) | accuracy (%): 72.42 -> 81.50 | Improved |
| add prospect-guided modeling (Pro-SF) vs rational classifier | Pro-SF · Spam (non-rational agents) | accuracy (%): 78.38 -> 89.34 | Improved |
| add prospect-guided modeling (Pro-SF) vs rational classifier | Pro-SF · Adult (mixed behavior agents) | accuracy (%): 74.31 -> 78.68 | Improved |
| use Pro-SF when agents are fully rational | Pro-SF · Adult (rational agents) | accuracy (%): 78.53 -> 75.31 | Hurt |
| use Pro-SF when agents are fully rational | Pro-SF · Credit (rational agents) | accuracy (%): 76.12 -> 77.02 | Improved |
| add loss aversion to reference+probability variant (ablation) | Pro-SF · Behavioral mechanism ablation | accuracy (%): 77.72 -> 78.92 | Improved |
| add probability distortion to reference+loss variant (ablation) | Pro-SF · Behavioral mechanism ablation | accuracy (%): 77.80 -> 78.92 | Improved |
| add reference bias to probability+loss variant (ablation) | Pro-SF · Behavioral mechanism ablation | accuracy (%): 77.45 -> 78.92 | Improved |
| add reference bias to probability-only variant (ablation) | Pro-SF · Behavioral mechanism ablation | over-defense error (%): 9.26 -> 6.28 | Improved |
| add probability distortion to Refe+Los variant (ablation) | Pro-SF · Behavioral mechanism ablation | under-defense error (%): 5.05 -> 3.22 | Improved |
| combine multiple mechanisms vs single mechanism (ablation) | Pro-SF · Behavioral mechanism ablation | accuracy (%): 73.50 -> 77.45 | Improved |
| vary kappa, gamma, alpha-beta and reference granularity (ablation) | Pro-SF · Parameter sensitivity | accuracy (%) | No clear effect |
Closest papers by meaning
- Understanding Model Selection for Learning in Strategic Environments (NeurIPS 2024 · similarity 0.86)
- Strategic Classification with Non-Linear Classifiers (NeurIPS 2025 · similarity 0.86)
- Should Decision-Makers Reveal Classifiers in Online Strategic Classification? (ICML 2025 · similarity 0.85)
- Who’s Gaming the System? A Causally-Motivated Approach for Detecting Strategic Adaptation (NeurIPS 2024 · similarity 0.85)
- Strategic Classification With Externalities (ICLR 2025 · similarity 0.85)
- Causal Strategic Learning with Competitive Selection (AAAI 2024 · 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.
