Multi-fidelity classification using Gaussian processes: accelerating the prediction of large-scale computational models

DOI10.1016/J.CMA.2019.112602MaRDI QIDQ2179219zbMATH OpenOpenAlexFDO

Authors Francisco Sahli Costabal, Paris Perdikaris, Daniel E. Hurtado, Ellen Kuhl

Publication date 12 May 2020

Published in Computer Methods in Applied Mechanics and Engineering (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1905.03406

zbMATH Keywords

Bayesian inferencemachine learningHamiltonian Monte Carlocardiac electrophysiologydata-driven modeling

Mathematics Subject Classification ID

Computational methods for problems pertaining to statistics (62-08) Classification and discrimination; cluster analysis (statistical aspects) (62H30)

Abstract: Machine learning techniques typically rely on large datasets to create accurate classifiers. However, there are situations when data is scarce and expensive to acquire. This is the case of studies that rely on state-of-the-art computational models which typically take days to run, thus hindering the potential of machine learning tools. In this work, we present a novel classifier that takes advantage of lower fidelity models and inexpensive approximations to predict the binary output of expensive computer simulations. We postulate an autoregressive model between the different levels of fidelity with Gaussian process priors. We adopt a fully Bayesian treatment for the hyper-parameters and use Markov Chain Mont Carlo samplers. We take advantage of the probabilistic nature of the classifier to implement active learning strategies. We also introduce a sparse approximation to enhance the ability of themulti-fidelity classifier to handle large datasets. We test these multi-fidelity classifiers against their single-fidelity counterpart with synthetic data, showing a median computational cost reduction of 23% for a target accuracy of 90%. In an application to cardiac electrophysiology, the multi-fidelity classifier achieves an F1 score, the harmonic mean of precision and recall, of 99.6% compared to 74.1% of a single-fidelity classifier when both are trained with 50 samples. In general, our results show that the multi-fidelity classifiers outperform their single-fidelity counterpart in terms of accuracy in all cases. We envision that this new tool will enable researchers to study classification problems that would otherwise be prohibitively expensive. Source code is available at https://github.com/fsahli/MFclass.

Recommendations

• Survey of multifidelity methods in uncertainty propagation, inference, and optimization
• Nonlinear information fusion algorithms for data-efficient multi-fidelity modelling
• Residual Gaussian process: a tractable nonparametric Bayesian emulator for multi-fidelity simulations
• A generalized probabilistic learning approach for multi-fidelity uncertainty quantification in complex physical simulations
• High-dimensional nonlinear multi-fidelity model with gradient-free active subspace method

Cites work

• scientific article; zbMATH DE number 1149423 (Why is no real title available?)
• scientific article; zbMATH DE number 2061729 (Why is no real title available?)
• scientific article; zbMATH DE number 1522714 (Why is no real title available?)
• A generalized multi-resolution expansion for uncertainty propagation with application to cardiovascular modeling
• A multi-resolution, non-parametric, Bayesian framework for identification of spatially-varying model parameters
• A unifying view of sparse approximate Gaussian process regression
• Approximations for binary Gaussian process classification
• Gaussian processes for machine learning.
• Large Sample Properties of Simulations Using Latin Hypercube Sampling
• Machine learning in drug development: characterizing the effect of 30 drugs on the QT interval using Gaussian process regression, sensitivity analysis, and uncertainty quantification
• Multi-fidelity Gaussian process regression for prediction of random fields
• Multifidelity Monte Carlo estimation of variance and sensitivity indices
• Multifidelity information fusion algorithms for high-dimensional systems and massive data sets
• Nonlinear information fusion algorithms for data-efficient multi-fidelity modelling
• On a Test of Whether one of Two Random Variables is Stochastically Larger than the Other
• Predicting the output from a complex computer code when fast approximations are available
• RECURSIVE CO-KRIGING MODEL FOR DESIGN OF COMPUTER EXPERIMENTS WITH MULTIPLE LEVELS OF FIDELITY
• Survey of multifidelity methods in uncertainty propagation, inference, and optimization
• The design and analysis of computer experiments.
• The no-U-turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo

Cited in

(11)

• Efficient estimation of cardiac conductivities: a proper generalized decomposition approach
• Gaussian process learning of nonlinear dynamics
• Editorial: Special issue on uncertainty quantification, machine learning, and data-driven modeling of biological systems
• Enhanced variable-fidelity surrogate-based optimization framework by Gaussian process regression and fuzzy clustering
• Stochastic multi-fidelity surrogate modeling of dendritic crystal growth
• Multifidelity Data Fusion via Gradient-Enhanced Gaussian Process Regression
• How drugs modulate the performance of the human heart
• MFclass
• Multi-fidelity Gaussian process regression for prediction of random fields
• Learning Enabled Constrained Black-Box Optimization
• A fast multi-fidelity method with uncertainty quantification for complex data correlations: application to vortex-induced vibrations of marine risers

Describes a project that uses

Uses Software

• GitHub

This page was built for publication: Multi-fidelity classification using Gaussian processes: accelerating the prediction of large-scale computational models

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2179219)