Research map / Probabilistic and scientific modeling / Variational and generative sampling
Tokenised Flow Matching for Hierarchical Simulation Based Inference
Giovanni Charles, Cosmo Nazzareno Santoni, Seth Flaxman, Elizaveta Semenova
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 likelihood-factorised single-site training instead of fully flattened NPE | TFMPE · Hierarchical SIR (n_s=50, N=5000) | l-C2ST: 1.91e-01 -> 1.21e-04 | Improved |
| use LF instead of posterior factorisation on Gaussian Mixture at n_s=50 | TFMPE · Hierarchical Gaussian Mixture (N=5000) | l-C2ST: 2.01e-01 -> 1.62e-03 | Improved |
| use LF instead of posterior factorisation on Two Moons | TFMPE · Hierarchical Two Moons (N=10,000) | l-C2ST: 5.69e-02 -> 1.16e-01 | Hurt |
| replace surrogate samples with true simulator samples (Direct) (ablation) | TFMPE with LF · Hierarchical Two Moons | calibration | Improved |
| swap LF for PF factorisation under TFMPE backbone (ablation) | TFMPE · Hierarchical Two Moons | calibration | No clear effect |
| add group embedding for site membership (ablation) | TFMPE · hierarchical benchmark tasks | l-C2ST | Improved |
| use MLP vector field instead of transformer (ablation) | TFMPE with LF · hierarchical benchmark tasks | l-C2ST | No clear effect |
| share one estimator for surrogate and posterior (Joint) (ablation) | TFMPE with LF · hierarchical benchmark tasks | stability | Hurt |
| use LF training on haemodynamics with 2 patients | TFMPE · Haemodynamics CFD (N=1000) | training cost (h): 6.6 h -> 3.3 h | Improved |
| draw observations from neural surrogate instead of CFD | TFMPE · Haemodynamics per-patient sampling | wall time: 11.88 s -> 6.41 ms | Improved |
| scale to n_s=100 sites with LF versus PF on Gaussian Mixture | TFMPE · Hierarchical Gaussian Mixture (n_s=100) | l-C2ST: 2.44e-01 -> 1.52e-02 | Improved |
| use LF instead of PF on SLCP at n_s=100 | TFMPE · Hierarchical SLCP (n_s=100) | l-C2ST: 1.33e-02 -> 2.75e-02 | Hurt |
Closest papers by meaning
- Consistency Models for Scalable and Fast Simulation-Based Inference (NeurIPS 2024 · similarity 0.89)
- Effortless, Simulation-Efficient Bayesian Inference using Tabular Foundation Models (NeurIPS 2025 · similarity 0.87)
- Inductive Domain Transfer In Misspecified Simulation-Based Inference (NeurIPS 2025 · similarity 0.86)
- Compositional simulation-based inference for time series (ICLR 2025 · similarity 0.86)
- Model-Informed Flows for Bayesian Inference (NeurIPS 2025 · similarity 0.86)
- 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.
