Estimating Higher-Order Moments Using Symmetric Tensor Decomposition
Abstract
We consider the problem of decomposing higher-order moment tensors, i.e., the sum of symmetric outer products of data vectors. Such a decomposition can be used to estimate the means in a Gaussian mixture model and for other applications in machine learning. The th-order empirical moment tensor of a set of observations of variables is a symmetric -way tensor. Our goal is to find a low-rank tensor approximation comprising symmetric outer products. The challenge is that forming the empirical moment tensors costs operations and storage, which may be prohibitively expensive; additionally, the algorithm to compute the low-rank approximation costs per iteration. Our contribution is avoiding formation of the moment tensor, computing the low-rank tensor approximation of the moment tensor implicitly using operations per iteration and no extra memory. This advance opens the door to more applications of higher-order moments since they can now be efficiently computed. We present numerical evidence of the computational savings and show an example of estimating the means for higher-order moments.
Cite
@article{arxiv.1911.03813,
title = {Estimating Higher-Order Moments Using Symmetric Tensor Decomposition},
author = {Samantha Sherman and Tamara G. Kolda},
journal= {arXiv preprint arXiv:1911.03813},
year = {2020}
}