Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Bayesian Optimization for Likelihood-Free Inference of Simulator-Based Statistical Models
Michael U. Gutmann, Jukka Corander
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 |
|---|---|---|---|
| model the discrepancy with Gaussian process regression and Bayesian optimization instead of sample averages | ABC likelihood approximation · Day care centers model | simulations per likelihood curve: 15,000 -> 50 | Improved |
| use BOLFI instead of MCMC with synthetic likelihood | synthetic likelihood MCMC · Ricker model | simulator calls: about 600 times fewer calls | Improved |
| use a log-Gaussian process instead of a Gaussian process for the discrepancy (ablation) | deterministic acquisition rule · Ricker model | approximation of limiting synthetic likelihood: better exploration and better approximation | Improved |
| use a stochastic instead of deterministic acquisition rule (ablation) | Bayesian optimization · Ricker model | approximation of limiting synthetic likelihood: more stable and more accurate approximations | Improved |
| use deterministic acquisition rule with noisy synthetic likelihood (N=50) (ablation) | Bayesian optimization · Ricker model | spatial exploration: very little spatial exploration at N=50 | Hurt |
| increase number of simulated data sets N from 50 to 50,000 (ablation) | synthetic likelihood · Ricker model | smoothness of likelihood curve: N = 50 to N = 50,000 smoother, still rugged | Improved |
Closest papers by meaning
- Learning Likelihood-Free Reference Priors (ICML 2025 · similarity 0.92)
- Simultaneous identification of models and parameters of scientific simulators (ICML 2024 · similarity 0.87)
- Effortless, Simulation-Efficient Bayesian Inference using Tabular Foundation Models (NeurIPS 2025 · similarity 0.87)
- Consistency Models for Scalable and Fast Simulation-Based Inference (NeurIPS 2024 · similarity 0.86)
- All-in-one simulation-based inference (ICML 2024 · similarity 0.86)
- Compositional simulation-based inference for time series (ICLR 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.
