D. Nguyen, E. Tsiligianni, and N. Deligiannis. (2018)cite arxiv:1812.01478Comment: in Proceedings of iTWIST'18, Paper-ID: 27, Marseille, France, November, 21-23, 2018.
Abstract
Matrix completion is one of the key problems in signal processing and machine
learning. In recent years, deep-learning-based models have achieved
state-of-the-art results in matrix completion. Nevertheless, they suffer from
two drawbacks: (i) they can not be extended easily to rows or columns unseen
during training; and (ii) their results are often degraded in case discrete
predictions are required. This paper addresses these two drawbacks by
presenting a deep matrix factorization model and a generic method to allow
joint training of the factorization model and the discretization operator.
Experiments on a real movie rating dataset show the efficacy of the proposed
models.
%0 Generic
%1 nguyen2018matrix
%A Nguyen, Duc Minh
%A Tsiligianni, Evaggelia
%A Deligiannis, Nikos
%D 2018
%K matrix_factorization
%T Matrix Factorization via Deep Learning
%U http://arxiv.org/abs/1812.01478
%V abs/1812.01478
%X Matrix completion is one of the key problems in signal processing and machine
learning. In recent years, deep-learning-based models have achieved
state-of-the-art results in matrix completion. Nevertheless, they suffer from
two drawbacks: (i) they can not be extended easily to rows or columns unseen
during training; and (ii) their results are often degraded in case discrete
predictions are required. This paper addresses these two drawbacks by
presenting a deep matrix factorization model and a generic method to allow
joint training of the factorization model and the discretization operator.
Experiments on a real movie rating dataset show the efficacy of the proposed
models.
@conference{nguyen2018matrix,
abstract = {Matrix completion is one of the key problems in signal processing and machine
learning. In recent years, deep-learning-based models have achieved
state-of-the-art results in matrix completion. Nevertheless, they suffer from
two drawbacks: (i) they can not be extended easily to rows or columns unseen
during training; and (ii) their results are often degraded in case discrete
predictions are required. This paper addresses these two drawbacks by
presenting a deep matrix factorization model and a generic method to allow
joint training of the factorization model and the discretization operator.
Experiments on a real movie rating dataset show the efficacy of the proposed
models.},
added-at = {2019-12-06T10:22:02.000+0100},
author = {Nguyen, Duc Minh and Tsiligianni, Evaggelia and Deligiannis, Nikos},
biburl = {https://www.bibsonomy.org/bibtex/28bd95ec345b13557eed78119f870ef4d/cyn7hia},
description = {Matrix Factorization via Deep Learning.pdf},
interhash = {f73b044271dd541d35b5286d9075e77a},
intrahash = {8bd95ec345b13557eed78119f870ef4d},
keywords = {matrix_factorization},
note = {cite arxiv:1812.01478Comment: in Proceedings of iTWIST'18, Paper-ID: 27, Marseille, France, November, 21-23, 2018},
timestamp = {2019-12-09T08:42:51.000+0100},
title = {Matrix Factorization via Deep Learning},
url = {http://arxiv.org/abs/1812.01478},
volume = {abs/1812.01478},
year = 2018
}