Research map / Learning theory and algorithms / Fairness
Tilted Empirical Risk Minimization
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 TERM instead of CRR | robust regression · Drug Discovery, 20% noise | test RMSE: 1.10 { (.07)} -> 1.08 { (.05)} | Helped |
| Use TERM instead of CRR | robust regression · Drug Discovery, 40% noise | test RMSE: 1.51 { (.08)} -> 1.10 { (.04)} | Helped |
| Use TERM instead of CRR | robust regression · Drug Discovery, 80% noise | test RMSE: 4.07 { (.06)} -> 1.68 { (.03)} | Helped |
| Use TERM instead of GCE | robust classification with Inception · CIFAR10, 20% noise | test accuracy: 0.805 { (.004)} -> 0.795 { (.004)} | Hurt |
| Use TERM instead of GCE | robust classification with Inception · CIFAR10, 40% noise | test accuracy: 0.750 { (.004)} -> 0.768 { (.004)} | Helped |
| Use TERM instead of GCE | robust classification with Inception · CIFAR10, 80% noise | test accuracy: 0.433 { (.005)} -> 0.455 { (.005)} | Helped |
| Apply sample-level and class-level hierarchical tilting | HIV-1 classification · 30% noise, 1:20 imbalance | overall test accuracy: 0.890 { (.009)} -> 0.900 { (.005)} | Helped |
| Apply class-level and annotator-level hierarchical tilting | HIV-1 classification · 30% noise, 1:20 imbalance | overall test accuracy: 0.890 { (.009)} -> 0.901 { (.010)} | Helped |
| Replace sample-level and class-level tilting with class-level and annotator-level tilting | hierarchical TERM on HIV-1 · 30% noise, 1:20 imbalance | rare-class test accuracy (Y=0): 0.778 { (.008)} -> 0.806 { (.020)} | Helped |
| Use TERM instead of RobustRegRisk | HIV-1 classification · clean data, 1:4 imbalance | overall test accuracy: 0.939 { (.004)} -> 0.937 { (.004)} | Hurt |
| Use TERM instead of FocalLoss | HIV-1 classification · clean data, 1:20 imbalance | overall test accuracy: 0.918 { (.003)} -> 0.921 { (.002)} | Helped |
| Use TERM instead of GCE | HIV-1 classification · 30% noise, 1:4 imbalance | overall test accuracy: 0.925 { (.005)} -> 0.924 { (.004)} | Hurt |
Closest papers by meaning
- Generalization and Robustness of the Tilted Empirical Risk (ICML 2025 · similarity 0.90)
- Differentially Private Empirical Risk Minimization with Smooth Non-Convex Loss Functions: A Non-Stationary View (AAAI 2019 · similarity 0.89)
- On the Asymptotic Distribution of the Minimum Empirical Risk (ICML 2024 · similarity 0.89)
- Efficient Algorithms for Empirical Group Distributionally Robust Optimization and Beyond (ICML 2024 · similarity 0.87)
- f-FERM: A Scalable Framework for Robust Fair Empirical Risk Minimization (ICLR 2024 · similarity 0.86)
- On the Efficiency of ERM in Feature Learning (NeurIPS 2024 · similarity 0.86)
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.