Deep composition of tensor-trains using squared inverse Rosenblatt transports
From MaRDI portal
Publication:2098237
Abstract: Characterising intractable high-dimensional random variables is one of the fundamental challenges in stochastic computation. The recent surge of transport maps offers a mathematical foundation and new insights for tackling this challenge by coupling intractable random variables with tractable reference random variables. This paper generalises the functional tensor-train approximation of the inverse Rosenblatt transport recently developed by Dolgov et al. (Stat Comput 30:603--625, 2020) to a wide class of high-dimensional non-negative functions, such as unnormalised probability density functions. First, we extend the inverse Rosenblatt transform to enable the transport to general reference measures other than the uniform measure. We develop an efficient procedure to compute this transport from a squared tensor-train decomposition which preserves the monotonicity. More crucially, we integrate the proposed order-preserving functional tensor-train transport into a nested variable transformation framework inspired by the layered structure of deep neural networks. The resulting deep inverse Rosenblatt transport significantly expands the capability of tensor approximations and transport maps to random variables with complicated nonlinear interactions and concentrated density functions. We demonstrate the efficiency of the proposed approach on a range of applications in statistical learning and uncertainty quantification, including parameter estimation for dynamical systems and inverse problems constrained by partial differential equations.
Recommendations
- Low-rank tensor reconstruction of concentrated densities with application to Bayesian inversion
- Approximation and sampling of multivariate probability distributions in the tensor train decomposition
- A distributed framework for the construction of transport maps
- Spectral tensor-train decomposition
- Sparse approximation of triangular transports. I: The finite-dimensional case
Cites work
- A continuous analogue of the tensor-train decomposition
- A Family of Nonparametric Density Estimation Algorithms
- A theory of pseudoskeleton approximations
- A transport-based multifidelity preconditioner for Markov chain Monte Carlo
- Alternating minimal energy methods for linear systems in higher dimensions
- An `empirical interpolation' method: Application to efficient reduced-basis discretization of partial differential equations
- Approximation and sampling of multivariate probability distributions in the tensor train decomposition
- Bayesian inference with optimal maps
- Chebyshev and Fourier spectral methods.
- Contributions to the theory of convex bodies
- CUR matrix decompositions for improved data analysis
- Data-driven model reduction for the Bayesian solution of inverse problems
- Data-driven optimal transport
- DMRG approach to fast linear algebra in the TT-format
- From Knothe's Transport to Brenier's Map and a Continuation Method for Optimal Transport
- High-dimensional integration: The quasi-Monte Carlo way
- How to find a good submatrix
- scientific article; zbMATH DE number 947416 (Why is no real title available?)
- Inference via low-dimensional couplings
- Likelihood-informed dimension reduction for nonlinear inverse problems
- Monte Carlo strategies in scientific computing
- Nonlinear model reduction via discrete empirical interpolation
- Optimal low-rank approximations of Bayesian linear inverse problems
- Pseudo-skeleton approximations by matrices of maximal volume
- Quasi-Monte Carlo and multilevel Monte Carlo methods for computing posterior expectations in elliptic inverse problems
- Rates of convergence of the Hastings and Metropolis algorithms
- Rectangular maximum-volume submatrices and their applications
- Remarks on a Multivariate Transformation
- Riemann manifold Langevin and Hamiltonian Monte Carlo methods. With discussion and authors' reply
- Scalable posterior approximations for large-scale Bayesian inverse problems via likelihood-informed parameter and state reduction
- Sequential Monte Carlo methods for high-dimensional inverse problems: a case study for the Navier-Stokes equations
- Simulating normalizing constants: From importance sampling to bridge sampling to path sampling
- Space and space-time modeling using process convolutions
- Sparse grids
- Spectral methods. Algorithms, analysis and applications.
- Spectral tensor-train decomposition
- Tensor Decompositions and Applications
- Tensor spaces and numerical tensor calculus
- Tensor-train decomposition
- The alternating linear scheme for tensor optimization in the tensor train format
- Transport map accelerated Markov chain Monte Carlo
- TT-cross approximation for multidimensional arrays
Cited in
(19)- Low-rank tensor reconstruction of concentrated densities with application to Bayesian inversion
- A unified performance analysis of likelihood-informed subspace methods
- Computation and learning in high dimensions. Abstracts from the workshop held August 1--7, 2021 (hybrid meeting)
- Tensor Train Neighborhood Preserving Embedding
- MIONet: Learning Multiple-Input Operators via Tensor Product
- Rank bounds for approximating Gaussian densities in the tensor-train format
- An efficient iterative method for reconstructing the refractive index in complex domains from far field data
- Scalable conditional deep inverse Rosenblatt transports using tensor trains and gradient-based dimension reduction
- Deep Importance Sampling Using Tensor Trains with Application to a Priori and a Posteriori Rare Events
- Mini-workshop: Nonlinear approximation of high-dimensional functions in scientific computing. Abstracts from the mini-workshop held October 15--20, 2023
- Variational Bayesian optimal experimental design with normalizing flows
- Reverse diffusion sampling with tensor train approximations of Hamilton-Jacobi-Bellman equations
- Uncertainty quantification. Abstracts from the workshop held April 20--25, 2025
- Data assimilation: from mathematical and statistical foundations to applications. Abstracts from the workshop held February 23--28, 2025
- An Eulerian approach to regularized JKO scheme with low-rank tensor decompositions for Bayesian inversion
- Measure transport via polynomial density surrogates
- An approximation theory framework for measure-transport sampling algorithms
- Tensor-Based Reduced-Order Modeling for Optimization-Based Inverse Problems
- Title not available (Why is no real title available?)
This page was built for publication: Deep composition of tensor-trains using squared inverse Rosenblatt transports
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2098237)
