Learning both hierarchical and temporal representation has been among the
long-standing challenges of recurrent neural networks. Multiscale recurrent
neural networks have been considered as a promising approach to resolve this
issue, yet there has been a lack of empirical evidence showing that this type
of models can actually capture the temporal dependencies by discovering the
latent hierarchical structure of the sequence. In this paper, we propose a
novel multiscale approach, called the hierarchical multiscale recurrent neural
networks, which can capture the latent hierarchical structure in the sequence
by encoding the temporal dependencies with different timescales using a novel
update mechanism. We show some evidence that our proposed multiscale
architecture can discover underlying hierarchical structure in the sequences
without using explicit boundary information. We evaluate our proposed model on
character-level language modelling and handwriting sequence modelling.
%0 Generic
%1 chung2016hierarchical
%A Chung, Junyoung
%A Ahn, Sungjin
%A Bengio, Yoshua
%D 2016
%K kallimachos longrange neuralnet rnn structure
%T Hierarchical Multiscale Recurrent Neural Networks
%U http://arxiv.org/abs/1609.01704
%X Learning both hierarchical and temporal representation has been among the
long-standing challenges of recurrent neural networks. Multiscale recurrent
neural networks have been considered as a promising approach to resolve this
issue, yet there has been a lack of empirical evidence showing that this type
of models can actually capture the temporal dependencies by discovering the
latent hierarchical structure of the sequence. In this paper, we propose a
novel multiscale approach, called the hierarchical multiscale recurrent neural
networks, which can capture the latent hierarchical structure in the sequence
by encoding the temporal dependencies with different timescales using a novel
update mechanism. We show some evidence that our proposed multiscale
architecture can discover underlying hierarchical structure in the sequences
without using explicit boundary information. We evaluate our proposed model on
character-level language modelling and handwriting sequence modelling.
@misc{chung2016hierarchical,
abstract = {Learning both hierarchical and temporal representation has been among the
long-standing challenges of recurrent neural networks. Multiscale recurrent
neural networks have been considered as a promising approach to resolve this
issue, yet there has been a lack of empirical evidence showing that this type
of models can actually capture the temporal dependencies by discovering the
latent hierarchical structure of the sequence. In this paper, we propose a
novel multiscale approach, called the hierarchical multiscale recurrent neural
networks, which can capture the latent hierarchical structure in the sequence
by encoding the temporal dependencies with different timescales using a novel
update mechanism. We show some evidence that our proposed multiscale
architecture can discover underlying hierarchical structure in the sequences
without using explicit boundary information. We evaluate our proposed model on
character-level language modelling and handwriting sequence modelling.},
added-at = {2018-03-07T11:36:13.000+0100},
author = {Chung, Junyoung and Ahn, Sungjin and Bengio, Yoshua},
biburl = {https://www.bibsonomy.org/bibtex/2e8dfe8680c7856a15bc3a87629b239ca/albinzehe},
description = {[1609.01704] Hierarchical Multiscale Recurrent Neural Networks},
interhash = {b80649c971bdac371d486f3f765a3263},
intrahash = {e8dfe8680c7856a15bc3a87629b239ca},
keywords = {kallimachos longrange neuralnet rnn structure},
note = {cite arxiv:1609.01704},
timestamp = {2018-03-07T11:36:13.000+0100},
title = {Hierarchical Multiscale Recurrent Neural Networks},
url = {http://arxiv.org/abs/1609.01704},
year = 2016
}