Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Truncated Marginal Neural Ratio Estimation
Benjamin Kurt Miller, Alex Cole, Patrick Forré, Gilles Louppe, Christoph Weniger
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 |
|---|---|---|---|
| Truncate prior (TMNRE) instead of untruncated MNRE with same simulation budget (ablation) | MNRE · 3-dim torus | C2ST-ddm / KL | Improved |
| Truncation cutoff epsilon scan from 1e-10 to 1e-1 (ablation) | TMNRE · 3-dim torus | C2ST-ddm per simulation: optimal 10^-6 | Mixed |
| epsilon 10^-6 versus global minimum 10^-4 (ablation) | TMNRE · 3-dim torus | C2ST-ddm per simulation: 10^-6 vs 10^-4 | No clear effect |
| Use TMNRE instead of rejection ABC and SMC-ABC | TMNRE · marginalized SBI benchmark | C2ST-ddm | Improved |
| Use TMNRE on wide-posterior tasks (Gaussian Linear, SLCP, SLCP Distractors) | TMNRE · marginalized SBI benchmark | C2ST-ddm | No clear effect |
| Estimate marginals directly (MNRE/TMNRE) instead of joint NRE/SNRE | NRE · 10-dim eggbox, 10k samples | posterior structure recovery | Improved |
| Marginal NRE versus joint NRE on CMB simulator | NRE · CMB 6-parameter, 5,000 simulations | corner plot accuracy | Improved |
Closest papers by meaning
- Compositional simulation-based inference for time series (ICLR 2025 · similarity 0.89)
- Effortless, Simulation-Efficient Bayesian Inference using Tabular Foundation Models (NeurIPS 2025 · similarity 0.89)
- Direct Amortized Likelihood Ratio Estimation (AAAI 2024 · similarity 0.88)
- Multilevel neural simulation-based inference (NeurIPS 2025 · similarity 0.88)
- All-in-one simulation-based inference (ICML 2024 · similarity 0.88)
- Simultaneous identification of models and parameters of scientific simulators (ICML 2024 · 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.
