Research map / Learning theory and algorithms / Prediction and decision losses
Beyond Individualized Recourse: Interpretable and Interactive Summaries of Actionable Recourses
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 AReS instead of the strongest listed baseline | DNN recourse generation · COMPAS | Recourse Accuracy: 99.40% -> 99.37% | No clear effect |
| Use AReS instead of the strongest listed baseline | DNN recourse generation · Credit | Recourse Accuracy: 50.32% -> 78.23% | Helped |
| Use AReS instead of the strongest listed baseline | DNN recourse generation · Bail | Recourse Accuracy: 92.23 % -> 96.81% | Helped |
| Use AReS instead of the lowest-cost listed baseline | DNN recourse generation · COMPAS, Credit, and Bail | Mean FCost: 3.42 -> 2.83; 2.48 -> 2.23; 2.90 -> 2.45 | Helped |
| Use AReS instead of the strongest listed baseline | RF recourse generation · COMPAS, Credit, and Bail | Recourse Accuracy: 65.88 % -> 72.43%; 31.32% -> 39.87%; 92.37% -> 97.11% | Helped |
| Use AReS instead of the lowest-cost listed baseline | RF recourse generation · COMPAS, Credit, and Bail | Mean FCost: 4.48 -> 2.52; 1.35 -> 1.09; 2.87 -> 1.78 | Helped |
| Use AReS instead of the strongest listed baseline | LR recourse generation · COMPAS and Credit | Recourse Accuracy: 100% -> 99.53%; 100% -> 99.61% | No clear effect |
| Use AReS instead of AR | LR recourse generation · Bail | Recourse Accuracy: 100% -> 100% | No clear effect |
| Use AReS instead of the lowest-cost listed baseline | LR recourse generation · COMPAS, Credit, and Bail | Mean FCost: 5.41 -> 4.02; 1.69 -> 1.28; 7.35 -> 6.45 | Helped |
| Present subgroup recourse summaries with AReS rather than averaged AR-LIME recourses | User study on a deliberately biased bail recourse model · Bias detection | Detection rate: 20% -> 90% | Helped |
| Present subgroup recourse summaries with AReS rather than averaged AR-LIME recourses | User study on a deliberately biased bail recourse model · Bias description | Correct description rate: 0% -> 70% | Helped |
| Use AReS rather than AR-LIME aggregates | User study with a biased 3-layer neural network · Bias detection | Detection rate: 44.4% -> 88.9% | Helped |
Closest papers by meaning
- Providing Fair Recourse over Plausible Groups (AAAI 2024 · similarity 0.86)
- Prediction without Preclusion: Recourse Verification with Reachable Sets (ICLR 2024 · similarity 0.86)
- Performative Validity of Recourse Explanations (NeurIPS 2025 · similarity 0.85)
- ElliCE: Efficient and Provably Robust Algorithmic Recourse via the Rashomon Sets (NeurIPS 2025 · similarity 0.85)
- From Search to Sampling: Generative Models for Robust Algorithmic Recourse (ICLR 2025 · similarity 0.85)
- Counterfactual Metarules for Local and Global Recourse (ICML 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.