Long Short-Term Memory (LSTM) recurrent neural networks (RNNs) outperform
traditional RNNs when dealing with sequences involving not only short-term
but also long-term dependencies. The decoupled extended Kalman filter
learning algorithm (DEKF) works well in online environments and reduces
significantly the number of training steps when compared to the standard
gradient-descent algorithms. Previous work on LSTM, however, has
always used a form of gradient descent and has not focused on true
online situations. Here we combine LSTM with DEKF and show that this
new hybrid improves upon the original learning algorithm when applied
to online processing.
%0 Conference Paper
%1 P'erez-Ortiz2002
%A Pérez-Ortiz, J.A.
%A Schmidhuber, J.
%A Gers, F.A.
%A Eck, D.
%B Artificial Neural Networks -- ICANN 2002 (Proceedings)
%C Berlin
%D 2002
%E Dorronsoro, J.
%I Springer
%K imported
%P 1055--1060
%T Improving Long-Term Online Prediction with Decoupled Extended Kalman
Filters
%X Long Short-Term Memory (LSTM) recurrent neural networks (RNNs) outperform
traditional RNNs when dealing with sequences involving not only short-term
but also long-term dependencies. The decoupled extended Kalman filter
learning algorithm (DEKF) works well in online environments and reduces
significantly the number of training steps when compared to the standard
gradient-descent algorithms. Previous work on LSTM, however, has
always used a form of gradient descent and has not focused on true
online situations. Here we combine LSTM with DEKF and show that this
new hybrid improves upon the original learning algorithm when applied
to online processing.
@inproceedings{P'erez-Ortiz2002,
abstract = {Long Short-Term Memory (LSTM) recurrent neural networks (RNNs) outperform
traditional RNNs when dealing with sequences involving not only short-term
but also long-term dependencies. The decoupled extended Kalman filter
learning algorithm (DEKF) works well in online environments and reduces
significantly the number of training steps when compared to the standard
gradient-descent algorithms. Previous work on LSTM, however, has
always used a form of gradient descent and has not focused on true
online situations. Here we combine LSTM with DEKF and show that this
new hybrid improves upon the original learning algorithm when applied
to online processing.},
added-at = {2010-02-27T01:05:18.000+0100},
address = {Berlin},
author = {P\'{e}rez-Ortiz, J.A. and Schmidhuber, J. and Gers, F.A. and Eck, D.},
biburl = {https://www.bibsonomy.org/bibtex/26187bf338e4d73998cb1bac525c6b541/tb2332},
booktitle = {{Artificial Neural Networks -- ICANN 2002 (Proceedings)}},
editor = {Dorronsoro, J.},
interhash = {62cb7d95fe98002e7fe3bdce9b23f58d},
intrahash = {6187bf338e4d73998cb1bac525c6b541},
keywords = {imported},
owner = {thierry},
pages = {1055--1060},
publisher = {Springer},
source = {OwnPublication},
timestamp = {2010-02-27T01:05:24.000+0100},
title = {Improving Long-Term Online Prediction with {Decoupled Extended Kalman
Filters}},
year = 2002
}