Iterative surrogate model optimization (ISMO): an active learning algorithm for PDE constrained optimization with deep neural networks
From MaRDI portal
Publication:2021252
Abstract: We present a novel active learning algorithm, termed as iterative surrogate model optimization (ISMO), for robust and efficient numerical approximation of PDE constrained optimization problems. This algorithm is based on deep neural networks and its key feature is the iterative selection of training data through a feedback loop between deep neural networks and any underlying standard optimization algorithm. Under suitable hypotheses, we show that the resulting optimizers converge exponentially fast (and with exponentially decaying variance), with respect to increasing number of training samples. Numerical examples for optimal control, parameter identification and shape optimization problems for PDEs are provided to validate the proposed theory and to illustrate that ISMO significantly outperforms a standard deep neural network based surrogate optimization algorithm.
Recommendations
- The deep Ritz method: a deep learning-based numerical algorithm for solving variational problems
- Isogeometric topology optimization based on deep learning
- Optimization with learning-informed differential equation constraints and its applications
- Physics-constrained deep learning for high-dimensional surrogate modeling and uncertainty quantification without labeled data
- Training a Neural-Network-Based Surrogate Model for Aerodynamic Optimisation Using a Gaussian Process
Cites work
- A machine learning framework for data driven acceleration of computations of differential equations
- Active learning
- Adaptive finite element methods for PDE-constrained optimal control problems
- Advanced Lectures on Machine Learning
- An artificial neural network as a troubled-cell indicator
- Applied shape optimization for fluids.
- Computational optimization of systems governed by partial differential equations
- Consistency of Lipschitz learning with infinite unlabeled data and finite labeled data
- Deep learning
- Deep learning observables in computational fluid dynamics
- Deep learning-based numerical methods for high-dimensional parabolic partial differential equations and backward stochastic differential equations
- Efficient shape optimization for certain and uncertain aerodynamic design
- Hidden physics models: machine learning of nonlinear partial differential equations
- scientific article; zbMATH DE number 5703572 (Why is no real title available?)
- scientific article; zbMATH DE number 1215244 (Why is no real title available?)
- scientific article; zbMATH DE number 1243473 (Why is no real title available?)
- scientific article; zbMATH DE number 2161998 (Why is no real title available?)
- Introduction to Genetic Algorithms
- Multidimensional variation for quasi-Monte Carlo
- On the mathematical foundations of learning
- Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
- Solving high-dimensional partial differential equations using deep learning
Cited in
(26)- A physics-informed multi-fidelity approach for the estimation of differential equations parameters in low-data or large-noise regimes
- Adaptive machine learning-based surrogate modeling to accelerate PDE-constrained optimization in enhanced oil recovery
- Error analysis for physics-informed neural networks (PINNs) approximating Kolmogorov PDEs
- Analysis of heterogeneous structures of non-separated scales using curved bridge nodes
- Learning finite element convergence with the multi-fidelity graph neural network
- An overview on deep learning-based approximation methods for partial differential equations
- A globally convergent method to accelerate large-scale optimization using on-the-fly model hyperreduction: application to shape optimization
- Higher-order quasi-Monte Carlo training of deep neural networks
- scientific article; zbMATH DE number 7626805 (Why is no real title available?)
- Modeling design and control problems involving neural network surrogates
- VI-DGP: a variational inference method with deep generative prior for solving high-dimensional inverse problems
- On the approximation of functions by tanh neural networks
- Learning high frequency data via the coupled frequency predictor-corrector triangular DNN
- On the spectral bias of coupled frequency predictor-corrector triangular DNN: the convergence analysis
- AONN: An Adjoint-Oriented Neural Network Method for All-At-Once Solutions of Parametric Optimal Control Problems
- wPINNs: Weak Physics Informed Neural Networks for Approximating Entropy Solutions of Hyperbolic Conservation Laws
- One-shot learning of surrogates in PDE-constrained optimization under uncertainty
- Error analysis for deep neural network approximations of parametric hyperbolic conservation laws
- The ADMM-PINNs algorithmic framework for nonsmooth PDE-constrained optimization: a deep learning approach
- Deep mixed residual method for solving PDE-constrained optimization problems
- Multi-fidelity learning of reduced order models for parabolic PDE constrained optimization
- Learning WENO for entropy stable schemes to solve conservation laws
- Multiscale wave-based identification of layer-specific geometric and viscoelastic parameters in heterogeneous multilayer composites using full-field measurements
- Neural network approaches for parameterized optimal control
- An operator learning perspective on parameter-to-observable maps
- Neural networks for Bayesian inverse problems governed by a nonlinear ODE
This page was built for publication: Iterative surrogate model optimization (ISMO): an active learning algorithm for PDE constrained optimization with deep neural networks
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2021252)
