Research map / Label-efficient and robust learning / Domain adaptation and generalization
Minimax-Optimal Aggregation for Density Ratio Estimation
Lukas Gruber, Markus Holzleitner, Sepp Hochreiter, Werner Zellinger
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 |
|---|---|---|---|
| aggregate models instead of cross-validation selection | KuLSIF · Geometric datasets (avg) | Bregman error: 11.205 -> 10.683 | Improved |
| aggregate models instead of cross-validation selection | Exp · Geometric datasets (avg) | Bregman error: 13.392 -> 11.529 | Improved |
| aggregate DeepLR models instead of CV-selected DeepLR (Ablation 1) (ablation) | DeepLR · Geometric c3,d1.70 | Bregman error: 7.280 -> 6.504 | Improved |
| tune KMM bandwidth by aggregation instead of median heuristic (Ablation 2) (ablation) | KMM · Geometric c3,d1.70 | Bregman error: 16.783 -> 13.223 | Improved |
| use minimax aggregation instead of model averaging (Ablation 3) (ablation) | KuLSIF · Geometric c3,d1.70 | Bregman error: 8.553 -> 8.320 | Improved |
| use aggregation instead of Super Learner (ablation) | KuLSIF · Geometric datasets (avg) | Bregman error: 11.009 -> 10.683 | Improved |
| use aggregation instead of Bayesian model averaging (ablation) | Exp · Geometric datasets (avg) | Bregman error: 12.972 -> 11.529 | Improved |
| aggregate instead of cross-validation | KuLSIF · MiniDomainNet | target accuracy: 0.528 -> 0.535 | Improved |
| aggregate instead of cross-validation | KuLSIF · HHAR | target accuracy: 0.737 -> 0.780 | Improved |
| aggregate instead of cross-validation | LR · Amazon Reviews | target accuracy: 0.785 -> 0.789 | Improved |
| increase number of models with suboptimal hyperparameters (Ablation 4) (ablation) | Aggregation · KuLSIF geometric | error | No clear effect |
| increase data dimensionality (Ablation 5) (ablation) | Cross-validation · Gaussian mixtures | error | Hurt |
Closest papers by meaning
- Projection Pursuit Density Ratio Estimation (ICML 2025 · similarity 0.88)
- Classification Diffusion Models: Revitalizing Density Ratio Estimation (NeurIPS 2024 · similarity 0.87)
- Overcoming Saturation in Density Ratio Estimation by Iterated Regularization (ICML 2024 · similarity 0.86)
- Bounds on $L_p$ Errors in Density Ratio Estimation via $f$-Divergence Loss Functions (ICLR 2025 · similarity 0.84)
- Binary Losses for Density Ratio Estimation (ICLR 2025 · similarity 0.84)
- Scaling Laws for Optimal Data Mixtures (NeurIPS 2025 · 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.
