Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Score Matched Neural Exponential Families for Likelihood-Free Inference
Lorenzo Pacchiardi, Ritabrata Dutta
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 |
|---|---|---|---|
| learn exponential family with score matching to recover exact sufficient statistics (weak MCC) | SM · Beta model | MCC weak out: 0.958 | Improved |
| compare strong vs weak identifiability of recovered statistics | SM · Beta model statistics | MCC: 0.958 -> 0.723 | Mixed |
| compare strong vs weak identifiability for natural parameters | SM · Gaussian natural parameters | MCC: 0.974 -> 0.972 | No clear effect |
| use SM-learned statistics in ABC (ABC-SM) instead of rejection-learned statistics (ABC-FP) | ABC · exponential family models | Wasserstein distance / RMSE | Improved |
| use SM-learned statistics in ABC instead of ABC-FP | ABC · AR(2) and MA(2) | Wasserstein distance / RMSE | Hurt |
| use exchange MCMC with learned likelihood (Exc-SM) instead of ABC-SM | Exc-SM · exponential family models | Wasserstein distance / RMSE | Hurt |
| use Exc-SM/Exc-SSM instead of PMC-RE/PMC-SL | Exc-SM · AR(2) and MA(2) | Wasserstein distance / RMSE | Improved |
| use ABC-SSM/Exc-SSM instead of ABC-FP | ABC-SSM · Lorenz96 (small and large) | Kernel and Energy scoring rules | Improved |
| use Exc-SSM (exchange MCMC) instead of ABC-SSM posteriors | Exc-SSM · Lorenz96 (small) | posterior width | Mixed |
| replace SM with sliced score matching (SSM) | SM · exponential family models | posterior accuracy | No clear effect |
Closest papers by meaning
- Fast ε-free Inference of Simulation Models with Bayesian Conditional Density Estimation (NeurIPS 2016 · similarity 0.89)
- Approximate Bayesian Computation with Domain Expert in the Loop (ICML 2022 · similarity 0.89)
- Sequential Neural Score Estimation: Likelihood-Free Inference with Conditional Score Based Diffusion Models (ICML 2024 · similarity 0.87)
- Effortless, Simulation-Efficient Bayesian Inference using Tabular Foundation Models (NeurIPS 2025 · similarity 0.87)
- Compositional simulation-based inference for time series (ICLR 2025 · similarity 0.87)
- Learning Likelihood-Free Reference Priors (ICML 2025 · similarity 0.87)
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.
