Journal of Computing and Information Science in Engineering

Recording of February 2023 Spotlight: Acceleration of a Physics-Based Machine Learning Approach for Modeling and Quantifying Model-Form Uncertainties and Performing Model Updating

Share This Post

The recording is now available for the February 28, 2023 spotlight talk by Professor Charbel Farhat (Stanford University) based on his JCISE paper: Marie-Jo Azzi, Chady Ghnatios, Philip Avery, and Charbel Farhat, “Acceleration of a Physics-Based Machine Learning Approach for Modeling and Quantifying Model-Form Uncertainties and Performing Model Updating,” ASME  J. Comput. Inf. Sci. Eng. Feb 2023, 23(1): 011009.


The nonparametric probabilistic method (NPM) for modeling and quantifying model-form uncertainties is a physics-based, computationally tractable, machine learning method for performing uncertainty quantification and model updating. It extracts from data information not captured by a deterministic, high-dimensional model (HDM) of dimension N and infuses it into a counterpart stochastic, hyperreduced, projection-based reduced-order model (SHPROM) of dimension n ≪ N. Here, the robustness and performance of NPM are improved using a two-pronged approach. First, the sensitivities of its stochastic loss function with respect to the hyperparameters are computed analytically, by tracking the complex web of operations underlying the construction of that function. Next, the theoretical number of hyperparameters is reduced from 𝒪(n2) to 𝒪(n), by developing a network of autoencoders that provides a nonlinear approximation of the dependence of the SHPROM on the hyperparameters. The robustness and performance of the enhanced NPM are demonstrated using two nonlinear, realistic, structural dynamics applications.


Visit the ASME Digital Collection archives for JCISE

More To Explore


Special Issue on Geometric Data Processing and Analysis for Advanced Manufacturing

Geometric information, such as three-dimensional (3D) shapes and network topologies, has been increasingly explored in manufacturing research. For example, characterizing geometric information in 3D-printed parts, in-situ or ex-situ, opens opportunities for defect detection, quality improvement, and product customization. However, geometric data mining remains critically challenging. Geometric information is embedded in complex data structures, such as 3D point clouds, graphs, meshes, voxels, high-dimensional images, and tensors, which possess challenges for analysis due to their high-dimensionality, high-volume, unstructured, multimodality characteristics. Additional challenges stem from compromised data quality (e.g., noisy and incomplete data), the need for registration, etc.


2023 Reviewer’s Recognition

The Editor and Editorial Board of the Journal of Computing and Information Science in Engineering would like to thank all of the reviewers for volunteering their expertise and time reviewing manuscripts in 2023. Serving as reviewers for the journal is a critical service necessary to maintain the quality of our publication and to provide the authors with a valuable peer review of their work.