Research map / Learning theory and algorithms / Prediction and decision losses
Towards Robust and Reliable Algorithmic Recourse
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 the minimax ROAR objective to generate recourses robust to model shifts | LR recourse generation · German credit (correction shift) | shifted-model validity: 0.54 ± 0.27 -> 0.94 ± 0.08 | Helped |
| Use the minimax ROAR objective to generate recourses robust to model shifts | LR recourse generation · SBA case (temporal shift) | shifted-model validity: 0.31 ± 0.09 -> 0.98 ± 0.02 | Helped |
| Use the minimax ROAR objective to generate recourses robust to model shifts | LR recourse generation · Student performance (geospatial shift) | shifted-model validity: 0.29 ± 0.09 -> 0.67 ± 0.19 | Helped |
| Add ROAR robustness objective to MINT | MINT recourse generation · German credit (correction shift) | shifted-model validity: 0.93 ± 0.07 -> 1.00 ± 0.00 | Helped |
| Use ROAR-LIME rather than CFE | NN recourse generation · German credit (correction shift) | shifted-model validity: 0.47 ± 0.06 -> 0.72 ± 0.10 | Helped |
| Use ROAR-LIME rather than CFE | NN recourse generation · SBA case (temporal shift) | shifted-model validity: 0.52 ± 0.09 -> 0.97 ± 0.02 | Helped |
| Use ROAR-LIME rather than CFE | NN recourse generation · Student performance (geospatial shift) | shifted-model validity: 0.48 ± 0.09 -> 0.97 ± 0.03 | Helped |
| Use ROAR rather than AR | LR recourse generation · German credit (correction shift) | average cost: 0.85 ± 0.14 -> 3.13 ± 0.32 | Hurt |
| Use ROAR rather than AR | LR recourse generation · German credit (correction shift) | average cost: 0.09 ± 0.02 -> 0.36 ± 0.08 | Hurt |
| Increase the magnitude of the data distribution shift | Recourse methods CFE, AR-LIME, and ROAR-LIME · Synthetic data (mean, variance, and combined shifts) | shifted-model validity | Hurt |
| Increase the magnitude of the data distribution shift | ROAR-LIME compared with CFE and AR-LIME · Synthetic data (mean, variance, and combined shifts) | shifted-model validity | Helped |
Closest papers by meaning
- From Search to Sampling: Generative Models for Robust Algorithmic Recourse (ICLR 2025 · similarity 0.88)
- SafeAR: Safe Algorithmic Recourse by Risk-Aware Policies (AAAI 2024 · similarity 0.88)
- ElliCE: Efficient and Provably Robust Algorithmic Recourse via the Rashomon Sets (NeurIPS 2025 · similarity 0.86)
- Algorithmic Recourse for Long-Term Improvement (ICML 2025 · similarity 0.86)
- Prediction without Preclusion: Recourse Verification with Reachable Sets (ICLR 2024 · similarity 0.86)
- Learning Decision Trees and Forests with Algorithmic Recourse (ICML 2024 · similarity 0.84)
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.