Legendre Decomposition for Tensors
Machine Learning
2020-01-29 v2 Machine Learning
Abstract
We present a novel nonnegative tensor decomposition method, called Legendre decomposition, which factorizes an input tensor into a multiplicative combination of parameters. Thanks to the well-developed theory of information geometry, the reconstructed tensor is unique and always minimizes the KL divergence from an input tensor. We empirically show that Legendre decomposition can more accurately reconstruct tensors than other nonnegative tensor decomposition methods.
Cite
@article{arxiv.1802.04502,
title = {Legendre Decomposition for Tensors},
author = {Mahito Sugiyama and Hiroyuki Nakahara and Koji Tsuda},
journal= {arXiv preprint arXiv:1802.04502},
year = {2020}
}
Comments
12 pages, 6 figures, accepted to the 32nd Annual Conference on Neural Information Processing Systems (NIPS 2018)