Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes
Multifidelity Simulation-based Inference for Computationally Expensive Simulators
Anastasia N. Krouglova, Hayden R Johnson, Basile Confavreux, Michael Deistler, Pedro J. Goncalves
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 |
|---|---|---|---|
| pre-train on low-fidelity simulations then fine-tune on high-fidelity | MF-NPE · four benchmark tasks (SIR, SLCP, OU process, Gaussian Blob) | C2ST/MMD/NLTP | Improved |
| increase number of low-fidelity pre-training simulations (ablation) | MF-NPE · benchmark tasks | posterior accuracy | Improved |
| increase low-fidelity samples from 10^4 to 10^5 (ablation) | MF-NPE · OU and SLCP | posterior accuracy | No clear effect |
| add multifidelity first round to TSNPE | MF-TSNPE · benchmark tasks | posterior accuracy | Improved |
| add acquisition function to MF-TSNPE (ablation) | MF-TSNPE-AF · OU, SLCP, SIR | posterior accuracy | Mixed |
| use MF-NPE instead of MF-ABC | MF-NPE · benchmark tasks | posterior accuracy | Improved |
| add parameters to high-fidelity model absent from low-fidelity model (ablation) | MF-NPE · OU process | posterior accuracy | Hurt |
| use MF-NPE instead of NPE | MF-NPE · multicompartment neuron model | computational cost: 4.44 ± 0.06 times lower | Improved |
| use MF-NPE instead of NPE | MF-NPE · recurrent spiking network | samples within firing rate bounds: almost 30% increase | Improved |
| perturb low-fidelity simulator with invertible coordinate reversal (information preserved) (ablation) | MF-NPE · OU2 transfer | posterior accuracy | Hurt |
Closest papers by meaning
- Effortless, Simulation-Efficient Bayesian Inference using Tabular Foundation Models (NeurIPS 2025 · similarity 0.88)
- All-in-one simulation-based inference (ICML 2024 · similarity 0.87)
- Multilevel neural simulation-based inference (NeurIPS 2025 · similarity 0.86)
- Simulation-Based Inference with Quantile Regression (ICML 2024 · similarity 0.86)
- Compositional simulation-based inference for time series (ICLR 2025 · similarity 0.86)
- Simultaneous identification of models and parameters of scientific simulators (ICML 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.
