Research map / Detection and segmentation / Open-set and few-shot detection
Learning Recourse on Instance Environment to Enhance Prediction Accuracy
Lokesh Nagalapatti, Guntakanti Sai Koushik, Abir De, Sunita Sarawagi
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 |
|---|---|---|---|
| train classifier with iterative greedy subsetting instead of full data (ablation) | RecourseNet classifier f_theta · Shapenet-Large | accuracy at 100% recourse: 71.93 ± 0.63 -> 77.14 ± 0.63 | Improved |
| train classifier with iterative greedy subsetting instead of full data (ablation) | RecourseNet classifier f_theta · Shapenet-Small | accuracy at 100% recourse: 62.97 ± 0.80 -> 74.13 ± 1.10 | Improved |
| train classifier with iterative greedy subsetting instead of full data (ablation) | RecourseNet classifier f_theta · Speech Commands | accuracy at 100% recourse: 51.85 ± 1.08 -> 65.76 ± 1.44 | Improved |
| train classifier with iterative greedy subsetting instead of full data (ablation) | RecourseNet classifier f_theta · Skin-Lesion | accuracy at 100% recourse: 56.42 ± 0.80 -> 68.62 ± 0.90 | Improved |
| iterative greedy subsetting instead of one-shot subsetting (ablation) | RecourseNet classifier f_theta · Shapenet-Small | accuracy at 100% recourse: 65.55 ± 1.11 -> 74.13 ± 1.10 | Improved |
| one-shot subsetting instead of full data (ablation) | classifier f_theta · Shapenet-Large | accuracy at 100% recourse: 71.93 ± 0.63 -> 72.63 ± 0.54 | No clear effect |
| one-shot subsetting instead of full data (ablation) | classifier f_theta · Skin-Lesion | accuracy at 100% recourse: 56.42 ± 0.80 -> 60.89 ± 1.11 | Improved |
| learned recourse trigger pi using expected gain from f_CF instead of score-based, loss-predictor or random triggers (ablation) | RecourseNet · all four datasets, varying budget b | accuracy after recourse | Improved |
| random trigger instead of learned trigger (ablation) | RecourseNet · all four datasets, varying budget b | accuracy after recourse | Hurt |
| add f_CF supervision term to recommender objective (7) vs objective (6) without it (ablation) | recourse recommender g_phi · Shapenet-Small | recourse accuracy | Improved |
| add f_CF supervision term to recommender objective (7) vs objective (6) without it (ablation) | recourse recommender g_phi · Shapenet-Large and Speech Commands | recourse accuracy | No clear effect |
| use full objective (7) with D minus D_delta handling instead of only-phi model (ablation) | recourse recommender g_phi · Shapenet-Large | recourse accuracy | Improved |
Closest papers by meaning
- From Search to Sampling: Generative Models for Robust Algorithmic Recourse (ICLR 2025 · similarity 0.83)
- Learning Decision Trees and Forests with Algorithmic Recourse (ICML 2024 · similarity 0.81)
- SafeAR: Safe Algorithmic Recourse by Risk-Aware Policies (AAAI 2024 · similarity 0.81)
- Prediction without Preclusion: Recourse Verification with Reachable Sets (ICLR 2024 · similarity 0.81)
- Reproduce, Replicate, Reevaluate. The Long but Safe Way to Extend Machine Learning Methods (AAAI 2024 · similarity 0.80)
- Providing Fair Recourse over Plausible Groups (AAAI 2024 · similarity 0.80)
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.
