Surrogate Modeling
Overview
Surrogate modeling studies how to approximate costly simulators or opaque predictors so they can be queried, optimized, or interpreted more efficiently. Common approaches include Gaussian processes and reduced-order models, neural operators and other deep networks, and hybrid or multi-fidelity methods that combine cheap simulations with a smaller number of accurate ones; active learning can select the most useful new evaluations. Applications range from PDE and climate emulation to engineering design, control, Bayesian optimization, and explanations of black-box models. A central challenge is balancing speed and average prediction accuracy with reliable uncertainty estimates, physical consistency, and performance on unseen regimes or decisions.
Latest papers
256 papers27 papers in the last four weeks, down 7% on the four weeks before. 0.2% of all new papers.
Shape-informed cardiac mechanics surrogates in data-scarce regimes via geometric encoding and generative augmentation
Interpolation-Driven Machine Learning Approaches for Plume Shine Dose Estimation: A Comparison of XGBoost, Random Forest, and TabNet
Position: A Dynamical Systems Perspective is Needed to Advance Time Series Modeling
Stabilizing Test-Time Adaptation of High-Dimensional Simulation Surrogates via D-Optimal Statistics
Adaptive Uncertainty-Guided Surrogates for Efficient phase field Modeling of Dendritic Solidification
Scaling laws for amplitude surrogates
Deep Jump Gaussian Processes for Surrogate Modeling of High-Dimensional Piecewise Continuous Functions
Surrogate Modeling and Explainable Artificial Intelligence for Complex Systems: A Workflow for Automated Simulation Exploration
Climate Surrogates for Scalable Multi-Agent Reinforcement Learning: A Case Study with CICERO-SCM
Surrogate Graph Partitioning for Spatial Prediction
Benchmarking Stochastic Interpolants for Modeling Physical Systems
Bifidelity Parameter Estimation Using Conditional Diffusion Models
Optimal Kernel Learning for Gaussian Process Models with High-Dimensional Input
active variables''. We propose an optimal kernel learning approach to identify these active variables, thereby overcoming GP model limitations and enhancing system understanding. Our method approximates the original GP model's covariance function through a convex combination of kernel functions, each utilizing low-dimensional subsets of input variables. Inspired by the Fedorov-Wynn algorithm from optimal design literature, we develop an optimal kernel learning algorithm to determine this approximation. We incorporate the effect heredity principle, a concept borrowed from the field of design and analysis of experiments'', to facilitate efficient computation. Through several examples, we demonstrate that the proposed method provides competitive prediction and variable selection, with favorable computational performance in the examples considered. It is an effective solution for interpreting the surrogate GP regression and simplifying the complex underlying system.