analiza aiqbench
This research evaluates uncertainty quantification methods for neural surrogates in computer-aided engineering, comparing Gaussian processes, Monte Carlo dropout, and deep ensembles across aerodynamics and crash dynamics datasets. The study finds that while all tested approaches generally correlate higher uncertainty with larger prediction errors, their relative performance varies significantly by dataset and evaluation metric. Consequently, the selection of specific UQ techniques must be tailored to the intended downstream engineering decisions rather than applying a one-size-fits-all solution.
Predictive Uncertainty for Neural CAE Surrogates
arXiv:2609.25430v1 Announce Type: cross Abstract: Neural surrogates can substantially accelerate computer-aided engineering (CAE) workflows, but their use in design requires uncertainty estimates that remain meaningful across varying geometries, spatial prediction fields, and engineering quantities of interest. We investigate how established uncertainty quantification (UQ) approaches behave when adapted to geometry-conditioned neural surrogates. We compare one closed-form and two sampling-based approaches-a Gaussian process (GP)-based method, concrete Monte Carlo (MC) dropout, and deep ensembles-and evaluate them on three large, industry-relevant CAE datasets for external aerodynamics and crash dynamics. We examine whether predicted uncertainties have credible magnitudes, identify locations with larger prediction errors, respond to unfamiliar inputs, and remain informative for derived engineering quantities. On the DrivAerStar dataset, where all three methods are compared, each generally assigns higher uncertainty to locations with larger prediction errors, and validation-based rescaling brings interval coverage close to nominal on a disjoint in-distribution test set. Results on AirFRANS and automotive crash also show useful error ranking and interval estimates, but the relative performance of the methods changes with the dataset and evaluation criterion. UQ methods and evaluation metrics should therefore be selected based on the intended downstream CAE decision.
Abstract:Neural surrogates can substantially accelerate computer-aided engineering (CAE) workflows, but their use in design requires uncertainty estimates that remain meaningful across varying geometries, spatial prediction fields, and engineering quantities of interest. We investigate how established uncertainty quantification (UQ) approaches behave when adapted to geometry-conditioned neural surrogates. We compare one closed-form and two sampling-based approaches-a Gaussian process (GP)-based method, concrete Monte Carlo (MC) dropout, and deep ensembles-and evaluate them on three large, industry-relevant CAE datasets for external aerodynamics and crash dynamics. We examine whether predicted uncertainties have credible magnitudes, identify locations with larger prediction errors, respond to unfamiliar inputs, and remain informative for derived engineering quantities. On the DrivAerStar dataset, where all three methods are compared, each generally assigns higher uncertainty to locations with larger prediction errors, and validation-based rescaling brings interval coverage close to nominal on a disjoint in-distribution test set. Results on AirFRANS and automotive crash also show useful error ranking and interval estimates, but the relative performance of the methods changes with the dataset and evaluation criterion. UQ methods and evaluation metrics should therefore be selected based on the intended downstream CAE decision.
Machine Learning (cs.LG); Artificial Intelligence (cs.AI); Probability (math.PR); Computational Physics (physics.comp-ph)
arXiv:2609.25430 [cs.LG]
(or arXiv:2609.25430v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2609.25430 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 287 cuvinte
Subiecte conexe
Uncertainty and Explainability in Deep Rough Volatility: A Neural Information-Theoretic Posterior Approach
arXiv · 28 sept. 2026
Modelling magnetic material properties with uncertainty-aware neural networks
arXiv · 6 oct. 2026
Conformal Prediction for Neural Operators: Distribution-Free Uncertainty Quantification in Physics Simulation
arXiv · 10 iun. 2026
Cycle-Consistent and Uncertainty-Aware Neural Surrogates for Tokamak Edge Plasmas
arXiv · 24 iul. 2026