DeepTensor: Low-Rank Tensor Decomposition with Deep Network Priors

by   Vishwanath Saragadam, et al.

DeepTensor is a computationally efficient framework for low-rank decomposition of matrices and tensors using deep generative networks. We decompose a tensor as the product of low-rank tensor factors (e.g., a matrix as the outer product of two vectors), where each low-rank tensor is generated by a deep network (DN) that is trained in a self-supervised manner to minimize the mean-squared approximation error. Our key observation is that the implicit regularization inherent in DNs enables them to capture nonlinear signal structures (e.g., manifolds) that are out of the reach of classical linear methods like the singular value decomposition (SVD) and principal component analysis (PCA). Furthermore, in contrast to the SVD and PCA, whose performance deteriorates when the tensor's entries deviate from additive white Gaussian noise, we demonstrate that the performance of DeepTensor is robust to a wide range of distributions. We validate that DeepTensor is a robust and computationally efficient drop-in replacement for the SVD, PCA, nonnegative matrix factorization (NMF), and similar decompositions by exploring a range of real-world applications, including hyperspectral image denoising, 3D MRI tomography, and image classification. In particular, DeepTensor offers a 6dB signal-to-noise ratio improvement over standard denoising methods for signals corrupted by Poisson noise and learns to decompose 3D tensors 60 times faster than a single DN equipped with 3D convolutions.


page 2

page 8

page 9


Heteroskedastic PCA: Algorithm, Optimality, and Applications

Principal component analysis (PCA) and singular value decomposition (SVD...

Reweighted Low-Rank Tensor Decomposition based on t-SVD and its Applications in Video Denoising

The t-SVD based Tensor Robust Principal Component Analysis (TRPCA) decom...

Multi-mode Core Tensor Factorization based Low-Rankness and Its Applications to Tensor Completion

Low-rank tensor completion has been widely used in computer vision and m...

Multilinear Low-Rank Tensors on Graphs & Applications

We propose a new framework for the analysis of low-rank tensors which li...

Deep Unfolded Tensor Robust PCA with Self-supervised Learning

Tensor robust principal component analysis (RPCA), which seeks to separa...

Tensor Denoising via Amplification and Stable Rank Methods

Tensors in the form of multilinear arrays are ubiquitous in data science...

H2TF for Hyperspectral Image Denoising: Where Hierarchical Nonlinear Transform Meets Hierarchical Matrix Factorization

Recently, tensor singular value decomposition (t-SVD) has emerged as a p...

Please sign up or login with your details

Forgot password? Click here to reset