Compositional Distributional Semantics with Long Short Term Memory
P. Le, and W. Zuidema. (2015)cite arxiv:1503.02510Comment: 10 pages, 7 figures.
Abstract
We are proposing an extension of the recursive neural network that makes use
of a variant of the long short-term memory architecture. The extension allows
information low in parse trees to be stored in a memory register (the `memory
cell') and used much later higher up in the parse tree. This provides a
solution to the vanishing gradient problem and allows the network to capture
long range dependencies. Experimental results show that our composition
outperformed the traditional neural-network composition on the Stanford
Sentiment Treebank.
Description
Compositional Distributional Semantics with Long Short Term Memory
%0 Generic
%1 le2015compositional
%A Le, Phong
%A Zuidema, Willem
%D 2015
%K lstm ma-zehe neuralnets rnn sentimentanalysis
%T Compositional Distributional Semantics with Long Short Term Memory
%U http://arxiv.org/abs/1503.02510
%X We are proposing an extension of the recursive neural network that makes use
of a variant of the long short-term memory architecture. The extension allows
information low in parse trees to be stored in a memory register (the `memory
cell') and used much later higher up in the parse tree. This provides a
solution to the vanishing gradient problem and allows the network to capture
long range dependencies. Experimental results show that our composition
outperformed the traditional neural-network composition on the Stanford
Sentiment Treebank.
@misc{le2015compositional,
abstract = {We are proposing an extension of the recursive neural network that makes use
of a variant of the long short-term memory architecture. The extension allows
information low in parse trees to be stored in a memory register (the `memory
cell') and used much later higher up in the parse tree. This provides a
solution to the vanishing gradient problem and allows the network to capture
long range dependencies. Experimental results show that our composition
outperformed the traditional neural-network composition on the Stanford
Sentiment Treebank.},
added-at = {2017-01-13T14:11:42.000+0100},
author = {Le, Phong and Zuidema, Willem},
biburl = {https://www.bibsonomy.org/bibtex/27cd995fcd0c9d70513c85e9a680412a1/albinzehe},
description = {Compositional Distributional Semantics with Long Short Term Memory},
interhash = {673a113cfe425f8eac7cf07e8a425cba},
intrahash = {7cd995fcd0c9d70513c85e9a680412a1},
keywords = {lstm ma-zehe neuralnets rnn sentimentanalysis},
note = {cite arxiv:1503.02510Comment: 10 pages, 7 figures},
timestamp = {2017-01-13T14:11:42.000+0100},
title = {Compositional Distributional Semantics with Long Short Term Memory},
url = {http://arxiv.org/abs/1503.02510},
year = 2015
}