GTM through time

C. M. Bishop; G. E. Hinton; I. G. D. Strachan

doi:10.1049/cp:19970711

GTM through time

C. M. Bishop, G. E. Hinton, I. G. D. Strachan

School of Informatics

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

Abstract

The standard GTM (generative topographic mapping) algorithm assumes that the data on which it is trained consists of independent, identically distributed (i.i.d.) vectors. For time series, however, the i.i.d. assumption is a poor approximation. In this paper we show how the GTM algorithm can be extended to model time series by incorporating it as the emission density in a hidden Markov model. Since GTM has discrete hidden states we are able to find a tractable EM algorithm, based on the forward-backward algorithm, to train the model. We illustrate the performance of GTM through time using flight recorder data from a helicopter

Original language	English
Title of host publication	Artificial Neural Networks, Fifth International Conference on (Conf. Publ. No. 440)
Publisher	Institute of Electrical and Electronics Engineers
Pages	111-116
Number of pages	6
ISBN (Print)	0-85296-690-3
DOIs	https://doi.org/10.1049/cp:19970711
Publication status	Published - Jul 1997

Access to Document

10.1049/cp:19970711

http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=607502

Cite this

@inproceedings{189a0ff9ed5f4632852befb3ab72a064,

title = "GTM through time",

abstract = "The standard GTM (generative topographic mapping) algorithm assumes that the data on which it is trained consists of independent, identically distributed (i.i.d.) vectors. For time series, however, the i.i.d. assumption is a poor approximation. In this paper we show how the GTM algorithm can be extended to model time series by incorporating it as the emission density in a hidden Markov model. Since GTM has discrete hidden states we are able to find a tractable EM algorithm, based on the forward-backward algorithm, to train the model. We illustrate the performance of GTM through time using flight recorder data from a helicopter",

author = "Bishop, \{C. M.\} and Hinton, \{G. E.\} and Strachan, \{I. G. D.\}",

year = "1997",

month = jul,

doi = "10.1049/cp:19970711",

language = "English",

isbn = "0-85296-690-3",

pages = "111--116",

booktitle = "Artificial Neural Networks, Fifth International Conference on (Conf. Publ. No. 440)",

publisher = "Institute of Electrical and Electronics Engineers",

address = "United States",

}

TY - GEN

T1 - GTM through time

AU - Bishop, C. M.

AU - Hinton, G. E.

AU - Strachan, I. G. D.

PY - 1997/7

Y1 - 1997/7

N2 - The standard GTM (generative topographic mapping) algorithm assumes that the data on which it is trained consists of independent, identically distributed (i.i.d.) vectors. For time series, however, the i.i.d. assumption is a poor approximation. In this paper we show how the GTM algorithm can be extended to model time series by incorporating it as the emission density in a hidden Markov model. Since GTM has discrete hidden states we are able to find a tractable EM algorithm, based on the forward-backward algorithm, to train the model. We illustrate the performance of GTM through time using flight recorder data from a helicopter

AB - The standard GTM (generative topographic mapping) algorithm assumes that the data on which it is trained consists of independent, identically distributed (i.i.d.) vectors. For time series, however, the i.i.d. assumption is a poor approximation. In this paper we show how the GTM algorithm can be extended to model time series by incorporating it as the emission density in a hidden Markov model. Since GTM has discrete hidden states we are able to find a tractable EM algorithm, based on the forward-backward algorithm, to train the model. We illustrate the performance of GTM through time using flight recorder data from a helicopter

U2 - 10.1049/cp:19970711

DO - 10.1049/cp:19970711

M3 - Conference contribution

SN - 0-85296-690-3

SP - 111

EP - 116

BT - Artificial Neural Networks, Fifth International Conference on (Conf. Publ. No. 440)

PB - Institute of Electrical and Electronics Engineers

ER -

GTM through time

Abstract

Access to Document

Fingerprint

Cite this