Research map / Probabilistic and scientific modeling / Variational and generative sampling
Estimation Beyond Data Reweighting: Kernel Method of Moments
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 KMM (MMD-based GEL with entropy regularization) instead of baselines for heteroskedastic IV | CMR parameter estimation · Heteroskedastic IV n=500 | parameter MSE: 0.40 vs 1.02 for FGEL | Helped |
| Use KMM instead of baselines for neural network IV regression | CMR neural network IV · Neural Network IV (abs, step, linear) | prediction MSE: slightly lower than best baseline on abs, step, linear | Helped |
| Use KMM instead of baselines for neural network IV regression | CMR neural network IV · Neural Network IV (sin) | prediction MSE: 0.88 vs 0.62 for FGEL | Hurt |
| Replace MMR with HSIC as validation metric for hyperparameter tuning and early stopping of variational estimators (ablation) | DeepGMM, FGEL, KMM · Heteroskedastic IV | prediction MSE: factor of 2-10 improvement | Helped |
| Choice of phi-divergence in entropy regularizer (KL vs others) (ablation) | KMM · CMR experiments | estimator quality | No clear effect |
| Lebesgue measure as reference distribution instead of KDE (ablation) | KMM reference distribution · CMR experiments | performance | Hurt |
| Empirical distribution as reference distribution instead of KDE (reweighting only) (ablation) | KMM reference distribution · CMR experiments | performance | Hurt |
| KDE-represented reference distribution (vs. empirical or Lebesgue) (ablation) | KMM reference distribution · CMR experiments | performance | Helped |
Closest papers by meaning
- The Polynomial Stein Discrepancy for Assessing Moment Convergence (ICML 2025 · similarity 0.86)
- Geometry-Aware Instrumental Variable Regression (ICML 2024 · similarity 0.85)
- Kernel conditional tests from learning-theoretic bounds (NeurIPS 2025 · similarity 0.85)
- Extracting Rare Dependence Patterns via Adaptive Sample Reweighting (ICML 2025 · similarity 0.84)
- On the Consistency of Kernel Methods with Dependent Observations (ICML 2024 · similarity 0.84)
- On the Asymptotic Distribution of the Minimum Empirical Risk (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.