Research map / Probabilistic and scientific modeling / Molecules and proteins
EDDI: Efficient Dynamic Discovery of High-Value Information with Partial VAE
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 EDDI-PNP instead of single-best variable ordering | EDDI on 6 UCI datasets · 6 UCI datasets | average ranking of AUIC: 5.20 (0.02) -> 5.08 (0.02 | Helped |
| use EDDI-PNP instead of single-best variable ordering | EDDI on MIMIC-III · MIMIC-III | average ranking on AUIC: 5.39 (0.01) -> 4.39 (0.01) | Helped |
| use EDDI-PNP instead of single-best variable ordering | EDDI on NHANES · NHANES | average ranking on AUIC: 5.17 (0.10) -> 4.80 (0.10) | Helped |
| use EDDI-PNP instead of LASSO-PNP | UCI active variable selection · 6 UCI datasets | average ranking of AUIC (RMSE-based): 4.41(0.02) -> 4.14(0.02) | Helped |
| use PointNet Plus instead of PointNet (ablation) | Partial VAE for MNIST inpainting · MNIST contiguous-region inpainting | Test ELBO (Reg.): -116.53 -> -113.19 | Helped |
| use PointNet Plus instead of zero-imputation VAE (ablation) | Partial VAE for MNIST inpainting · MNIST random missing pixels | Test ELBO (Rnd.): -116.01 -> -114.01 | Helped |
| add an observation mask to zero-imputation VAE (ablation) | Partial VAE for MNIST inpainting · MNIST random missing pixels | Test ELBO (Rnd.): -116.01 -> -118.61 | Hurt |
| use PointNet Plus instead of PointNet (ablation) | EDDI on UCI datasets · 6 UCI datasets | average ranking of AUIC: 5.25 (0.02) -> 5.08 (0.02 | Helped |
| use PointNet Plus instead of zero-imputation (ablation) | EDDI on UCI datasets · 6 UCI datasets | average ranking of AUIC: 5.72 (0.03) -> 5.08 (0.02 | Helped |
| use PointNet Plus instead of masked zero-imputation (ablation) | EDDI on MIMIC-III · MIMIC-III | average ranking on AUIC: 4.91 (0.01) -> 4.39 (0.01) | Helped |
| use PointNet Plus instead of zero-imputation (ablation) | EDDI on MIMIC-III · MIMIC-III | average ranking on AUIC: 8.83 (0.01) -> 4.39 (0.01) | Helped |
| use PointNet Plus instead of PointNet (ablation) | EDDI on NHANES · NHANES | average ranking on AUIC: 5.28 (0.10) -> 4.80 (0.10) | Helped |
Closest papers by meaning
- Coherent Hierarchical Multi-Label Learning to Defer for Medical Imaging (NeurIPS 2026 · similarity 0.75)
- Learning with Selectively Labeled Data from Multiple Decision-makers (ICML 2025 · similarity 0.75)
- ED-Copilot: Reduce Emergency Department Wait Time with Language Model Diagnostic Assistance (ICML 2024 · similarity 0.75)
- Treatment Effect Estimation for Optimal Decision-Making (NeurIPS 2025 · similarity 0.74)
- Stochastic Encodings for Active Feature Acquisition (ICML 2025 · similarity 0.74)
- Decision Theoretic Foundations for Conformal Prediction: Optimal Uncertainty Quantification for Risk-Averse Agents (ICML 2025 · similarity 0.74)
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.