Hybrid Neural Simulation-Based Inference for Robust Applications and Limited-Budget Scenarios
Abstract
We develop two hybrid techniques that approach the performance of neural simulation-based inference (NSBI) analyses while substantially reducing the computational cost of inference and preserving some or all of the reliability guarantees of parametric methods. The first approach is broadly applicable, while the second is tailored to a class of particle physics analyses that admit a semi-parametric NSBI formulation. With only a modest compromise in raw sensitivity, these methods represent an important step toward computationally efficient NSBI in offline analyses and also open the door to the exploration of trigger-level applications in the future. Based on our comparison studies, we recommend the use of our first approach, Latent Categories, for robust and efficient inference.
Related
- Simulation-Based Inference and Unbinned Asimov Construction with Hybrid Neural Density Estimation
- SCALE: Simulation-Calibrated Amortized Learning for Energy Materials (A hybrid architecture connecting deterministic modeling, real-world data, and transformer-scale inference for accelerated energy-materials discovery)
- Quantum-Correlation-Based Multi-Emitter Localization in Information-Limited Regimes via Learned Inference
- Optimizing diffusion-limited transport, with applications to electrochemical systems
- Spectroscopy of analogue black holes using simulation-based inference
BibTeX
@misc{benevedes2026hybrid,
title = {{Hybrid Neural Simulation-Based Inference for Robust Applications and Limited-Budget Scenarios}},
author = {Sean Benevedes and Mani Dehghan and Aishik Ghosh and Tae Hyoun Park},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.36044},
eprint = {2609.36044},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.36044},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Hybrid Neural Simulation-Based Inference for Robust Applications and Limited-Budget Scenarios
AU - Sean Benevedes
AU - Mani Dehghan
AU - Aishik Ghosh
AU - Tae Hyoun Park
PY - 2026
DA - 2026/09/28
PB - arXiv
DO - 10.48550/arxiv.2609.36044
UR - https://arxiv.org/abs/2609.36044
AB - We develop two hybrid techniques that approach the performance of neural simulation-based inference (NSBI) analyses while substantially reducing the computational cost of inference and preserving some or all of the reliability guarantees of parametric methods. The first approach is broadly applicable, while the second is tailored to a class of particle physics analyses that admit a semi-parametric NSBI formulation. With only a modest compromise in raw sensitivity, these methods represent an important step toward computationally efficient NSBI in offline analyses and also open the door to the exploration of trigger-level applications in the future. Based on our comparison studies, we recommend the use of our first approach, Latent Categories, for robust and efficient inference.
ER -