Generative modeling via tensor train sketching

2022/02/23 by YH. Hur, J. G. Hoskins, Hur, YH. +7 · 4 citations
Computer Science · Mathematics · #15A69 #62Gxx #FOS: Computer and information sciences #FOS: Mathematics #Machine Learning (cs.LG) #Machine Learning and Algorithms #Markov Chains and Monte Carlo Methods #Numerical Analysis (math.NA) #Tensor decomposition and applications

paper · pdf · doi:10.48550/arxiv.2202.11788

openalex publication_date 2022/02/23 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/01

Abstract

In this paper, we introduce a sketching algorithm for constructing a tensor train representation of a probability density from its samples. Our method deviates from the standard recursive SVD-based procedure for constructing a tensor train. Instead, we formulate and solve a sequence of small linear systems for the individual tensor train cores. This approach can avoid the curse of dimensionality that threatens both the algorithmic and sample complexities of the recovery problem. Specifically, for Markov models under natural conditions, we prove that the tensor cores can be recovered with a sample complexity that scales logarithmically in the dimensionality. Finally, we illustrate the performance of the method with several numerical experiments.

Cited by

• Optimal Bias-variance Tradeoff in Matrix and Tensor Estimation
• Variational inference and density estimation with non-negative tensor train
• Accelerated Inchworm Method with Tensor-Train Bath Influence Functional
• Inchworm tensor train hybridization expansion quantum impurity solver

Related

• Generative Modeling via Hierarchical Tensor Sketching
• Generative Modeling via Tree Tensor Network States
• Streaming Tensor Train Approximation
• High-dimensional density estimation with tensorizing flow
• Plastic tensor networks for interpretable generative modeling