Research map / Probabilistic and scientific modeling / Variational and generative sampling
Target–Aware Bayesian Inference: How to Beat Optimal Conventional Estimators
Tom Rainforth, Adam Golinski, Frank Wood, Sheheryar Zaidi
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 |
|---|---|---|---|
| use target-aware TAAIS instead of SNIS-AIS targeting p(x,y) | adaptive importance sampling · Gaussian example, 10 and 25 dimensions | log relative squared error: -12.85 -> -22.25 (D=10, y=2) | Improved |
| use TAAIS instead of SNIS-AIS targeting p(x,y) at D=50, y=2 | adaptive importance sampling · Gaussian example, D=50, y=2 | log relative squared error: -7.37 -> -6.91 | No clear effect |
| use TAAIS instead of SNIS-AIS targeting p(x,y)f(x) | adaptive importance sampling · Gaussian example, D=10, y=5 | log relative squared error: 5.31 -> -21.21 | Improved |
| increase separation y from 2 to 5 for SNIS-AIS targeting p(x,y) (ablation) | SNIS-AIS gamma2 · Gaussian example, D=10 | log relative squared error: -12.85 -> -1.58 | Hurt |
| increase separation y from 2 to 5 for TAAIS (ablation) | TAAIS · Gaussian example, D=10 | log relative squared error: -22.25 -> -21.21 | No clear effect |
| use TABI instead of optimal SNIS estimator | Gamma-Gaussian toy example · Figure 1 example | relative squared error: errors around two orders of magnitude less than the optimal SNIS sampler | Improved |
| use TAAIS instead of the optimal SNIS sampler | adaptive importance sampling · Gaussian example, 10 and 25 dimensions | relative squared error: outperformed the optimal SNIS sampler within the allocated budget | Improved |
| use TAAIS instead of the optimal SNIS sampler when proposal family cannot represent the targets | PI-MAIS adaptive sampler · Banana example | relative squared error | No clear effect |
| use TAAIS instead of SNIS-AIS baselines and MCMC | PI-MAIS adaptive sampler · Banana example | relative squared error: many orders of magnitude lower error than all the baseline approaches | Improved |
Closest papers by meaning
- Amortized Monte Carlo Integration (ICML 2019 · similarity 0.89)
- Effortless, Simulation-Efficient Bayesian Inference using Tabular Foundation Models (NeurIPS 2025 · similarity 0.88)
- Amortized Simulation-Based Inference in Generalized Bayes via Neural Posterior Estimation (ICML 2026 · similarity 0.86)
- Consistency Models for Scalable and Fast Simulation-Based Inference (NeurIPS 2024 · similarity 0.85)
- Leveraging Self-Consistency for Data-Efficient Amortized Bayesian Inference (ICML 2024 · similarity 0.85)
- Learning Likelihood-Free Reference Priors (ICML 2025 · similarity 0.85)
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.
