Research map / Probabilistic and scientific modeling / Variational and generative sampling
Conditional Distribution Function Estimation Using Neural Networks for Censored and Uncensored Data
Bingqing Hu, Bin Nan
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 |
|---|---|---|---|
| Replace L2-loss NN with full-likelihood NN for conditional distribution | NN conditional distribution estimator · Simulation Setup 2 uncensored, n=1000, c=0.5 | median squared error: 0.09 -> 0.02 | Improved |
| Replace L2-loss NN with full-likelihood NN | NN estimator · Simulation Setup 2 uncensored, n=5000, c=1 | median squared error: 0.13 -> 0.05 | Improved |
| Replace L2-loss NN with full-likelihood NN | NN estimator · Simulation Setup 2 uncensored, n=5000, c=1 | mean squared error: 6.47 -> 6.34 | Improved |
| Replace L2-loss NN with full-likelihood NN when errors are uncorrelated with constant variance | NN estimator · Simulation Setup 1, n=5000, c=1 | mean squared error: 1.73 -> 1.73 | No clear effect |
| Replace L2-loss NN with full-likelihood NN | NN estimator · Simulation Setup 1, n=1000, c=0.5 | median squared error: 0.17 -> 0.16 | Improved |
| Increase sample size from 1000 to 5000 | Proposed NN, Setup 2, c=0.5 · Simulation Setup 2 | 90% coverage rate: 0.87 -> 0.91 | Improved |
| Increase noise c from 0.5 to 1 | Proposed NN, Setup 1 n=1000 · Simulation Setup 1 | mean squared error: 0.51 -> 1.84 | Hurt |
| Use full-likelihood NN instead of Cox-Time | Cox-Time · SUPPORT | IBS: 0.212 -> 0.195 | Improved |
| Use full-likelihood NN instead of Cox-Time | Cox-Time · METABRIC | IBLL: -0.511 -> -0.496 | Improved |
| Use full-likelihood NN instead of Cox-Time | Cox-Time · Rot.&GBSG | IBS: 0.169 -> 0.176 | Hurt |
| Use full-likelihood NN instead of Cox-Time | Cox-Time · Rot.&GBSG | C-index: 0.677 -> 0.680 | Improved |
| Use full-likelihood NN instead of Cox-Time | Cox-Time · SUPPORT | C-index: 0.629 -> 0.609 | Hurt |
Closest papers by meaning
- A Unified View on Learning Unnormalized Distributions via Noise-Contrastive Estimation (ICML 2025 · similarity 0.80)
- Neural Conditional Probability for Uncertainty Quantification (NeurIPS 2024 · similarity 0.80)
- Generative Conditional Distributions by Neural (Entropic) Optimal Transport (ICML 2024 · similarity 0.80)
- Conditional Distribution Compression via the Kernel Conditional Mean Embedding (NeurIPS 2025 · similarity 0.79)
- A Likelihood Based Approach to Distribution Regression Using Conditional Deep Generative Models (ICML 2025 · similarity 0.79)
- Kernel conditional tests from learning-theoretic bounds (NeurIPS 2025 · similarity 0.78)
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.
