Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Truncated proposals for scalable and hassle-free simulation-based inference
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 truncated-prior proposals for sequential posterior estimation instead of APT's proposal and modified loss | TSNPE versus APT · Six benchmark tasks | C2ST accuracy | No clear effect |
| Use sequential inference rather than prior-only NPE | TSNPE versus NPE · Six benchmark tasks | C2ST accuracy | Helped |
| Use TSNPE rather than APT | Inference on Gaussian Linear and SLCP · Gaussian Linear and SLCP | C2ST accuracy | Hurt |
| Use TSNPE rather than APT | Inference on SIR and Lotka-Volterra · SIR and Lotka-Volterra | C2ST accuracy | Helped |
| Apply truncated-prior proposals and maximum-likelihood training rather than APT | Pyloric network inference · Pyloric network | posterior inference feasibility: 0.02% after the second round; 0.0000% after the third round | Helped |
| Apply truncated-prior proposals and maximum-likelihood training rather than APT | L5PC inference · L5PC | posterior inference feasibility: none of the 10M | Helped |
| Use an ensemble of neural density estimators (ablation) | TSNPE · L5PC | expected coverage | Helped |
| Increase the SIR oversampling factor rather than use too low a value (ablation) | TSNPE with SIR proposal sampling · General | posterior approximation accuracy: K=1024 | Helped |
| Use a larger truncation level epsilon (ablation) | TSNPE · Posterior tails | posterior tail accuracy: epsilon = 0.1 | Hurt |
| Apply truncated proposals to APT (ablation) | APT · Benchmark tasks | C2ST accuracy | Mixed |
Closest papers by meaning
- Sequential Neural Score Estimation: Likelihood-Free Inference with Conditional Score Based Diffusion Models (ICML 2024 · similarity 0.89)
- Effortless, Simulation-Efficient Bayesian Inference using Tabular Foundation Models (NeurIPS 2025 · similarity 0.89)
- Active Sequential Posterior Estimation for Sample-Efficient Simulation-Based Inference (NeurIPS 2024 · similarity 0.88)
- Simulation-Based Inference with Quantile Regression (ICML 2024 · similarity 0.87)
- Simultaneous identification of models and parameters of scientific simulators (ICML 2024 · similarity 0.86)
- Multilevel neural simulation-based inference (NeurIPS 2025 · similarity 0.86)
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.