Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Predictive Approximate Bayesian Computation via Saddle Points
Yingxiang Yang, Bo Dai, Negar Kiyavash, Niao He
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 P-ABC instead of K2-ABC | ABC parameter estimation · Synthetic I, d=16, M=10 | MSE: 1.283 -> [0.182, 0.155] | Improved |
| use P-ABC instead of DR-ABC | ABC parameter estimation · Synthetic I, d=256, M=10 | MSE: 21.324 -> [4.266,1.399] | Improved |
| use P-ABC instead of DR-ABC | ABC parameter estimation · Synthetic I, d=1, M=10 | MSE: 0.083 -> [0.009,0.010] | Improved |
| P-ABC versus the optimal closed-form estimator | P-ABC · Synthetic I, d=256 | MSE: 0.818 -> [4.266,1.399] | Hurt |
| P-ABC versus the average estimator at low dimension | P-ABC · Synthetic I, d=1 | MSE: 0.008 -> [0.009,0.010] | No clear effect |
| P-ABC versus theoretically optimal estimator in scalar case | P-ABC · Synthetic I d=1 M=1 | MSE: 0.0411 -> 0.0416 | No clear effect |
| use P-ABC instead of EP-ABC | ABC · Synthetic II Gaussian mixture | MSE: 0.06 -> 0.004 | Improved |
| use P-ABC instead of EP-ABC | ABC · Synthetic II Gaussian mixture | sets of samples per minute: 10 -> 200 | Improved |
| use P-ABC instead of K2-ABC | ABC · Ecological dynamic system | MSE: 24.7 -> 12.9 | Improved |
| use P-ABC instead of DR-ABC | ABC · Ecological dynamic system | MSE: 16.4 -> 12.9 | Improved |
| train P-ABC with a mismatched (skewed) prior, increasing a (ablation) | P-ABC · Synthetic II prior shift | estimate range around true parameter | Hurt |
Closest papers by meaning
- Approximate Bayesian Computation with Domain Expert in the Loop (ICML 2022 · similarity 0.90)
- DR-ABC: Approximate Bayesian Computation with Kernel-Based Distribution Regression (ICML 2016 · similarity 0.90)
- Approximate Bayesian Computation via Classification (JMLR 2022 · similarity 0.87)
- Fast ε-free Inference of Simulation Models with Bayesian Conditional Density Estimation (NeurIPS 2016 · similarity 0.87)
- Scalable Bayesian Learning with posteriors (ICLR 2025 · similarity 0.85)
- Practical Bayesian Algorithm Execution via Posterior Sampling (NeurIPS 2024 · 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.
