Research map / Learning theory and algorithms / Sample complexity
The AIM and EM Algorithms for Learning from Coarse Data
Manfred Jaeger
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 |
|---|---|---|---|
| use AIM instead of EM | EM (1d-b, B1 binning) · 1d Gaussian binned data, highly non-CAR coarsening (t1, t2, c1) | RMSE of parameter estimates | Improved |
| use AIM instead of EM | EM (1d-b, B1 binning) · 1d Gaussian binned data, nearly CAR coarsening (t3, t4, c2) | RMSE of parameter estimates | Hurt |
| initialize AIM from EM estimates (EM-AIM) instead of from ACA (ablation) | AIM · 1d Gaussian binned data, B1, t1 and t2 | RMSE of parameter estimates | Hurt |
| run AIM iterations after EM (EM-AIM) (ablation) | EM · 1d Gaussian binned data, B1, t1 and t2 | RMSE of parameter estimates | Improved |
| run AIM iterations after EM (EM-AIM) (ablation) | EM · 1d-b, B2 binning, N >= 10^4 | RMSE of parameter estimates | Improved |
| use AIM instead of EM | EM (1d-b, B2 binning) · 1d-b, B2 binning, non-CAR t1 and t2 with large samples | RMSE of parameter estimates | Improved |
| use AIM instead of EM | EM (1d-b, B2 binning) · 1d-b, B2 binning, non-CAR c1 | RMSE of parameter estimates | Hurt |
| use EM with B2 binning instead of ACA | ACA estimate · 1d-b, B2 binning | RMSE of parameter estimates | Improved |
| increase sample size to N = 1m with automatic granularity selection (ablation) | AIM · 1d-b, N = 1,000,000 | RMSE of parameter estimates | Hurt |
| make coarsening more nearly CAR (t1 to t4, c1 to c2) (ablation) | AIM · 1d-b, B1 binning | RMSE of parameter estimates | Hurt |
| minimize KL score across restarts for solution selection (ablation) | AIM · 1d-b, central coarsening c1 vs c2, N = 1m, 50 restarts | correlation of KL score with error | Mixed |
Closest papers by meaning
- Distribution Shift in Missing Data Imputation: A Risk-Based Perspective and Importance-Weighted Correction under MAR (NeurIPS 2026 · similarity 0.88)
- Random features models: a way to study the success of naive imputation (ICML 2024 · similarity 0.86)
- Diffusion Transformers for Imputation: Statistical Efficiency and Uncertainty Quantification (NeurIPS 2025 · similarity 0.85)
- Generative Modeling under Non-Monotone MAR Missingness via Approximate Wasserstein Gradient Flows (NeurIPS 2026 · similarity 0.85)
- Identification and Estimation for Nonignorable Missing Data: A Data Fusion Approach (ICML 2024 · similarity 0.85)
- Parameter Estimation in DAGs from Incomplete Data via Optimal Transport (ICML 2024 · similarity 0.85)
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.
