MCMC for Hierarchical Semi-Markov Conditional Random fields

Deep architecture such as hierarchical semi-Markov models is an important class of models for nested sequential data. Current exact inference schemes either cost cubic time in sequence length, or exponential time in model depth. These costs are prohibitive for large-scale problems with arbitrary len...

Full description

Bibliographic Details
Main Authors: Truyen, Tran, Phung, Dinh, Venkatesh, Svetha, Bui, Hung H.
Other Authors: Li Deng
Format: Conference Paper
Published: unknown 2009
Online Access:http://hdl.handle.net/20.500.11937/42222
_version_ 1848756360584364032
author Truyen, Tran
Phung, Dinh
Venkatesh, Svetha
Bui, Hung H.
author2 Li Deng
author_facet Li Deng
Truyen, Tran
Phung, Dinh
Venkatesh, Svetha
Bui, Hung H.
author_sort Truyen, Tran
building Curtin Institutional Repository
collection Online Access
description Deep architecture such as hierarchical semi-Markov models is an important class of models for nested sequential data. Current exact inference schemes either cost cubic time in sequence length, or exponential time in model depth. These costs are prohibitive for large-scale problems with arbitrary length and depth. In this contribution, we propose a new approximation technique that may have the potential to achieve sub-cubic time complexity in length and linear time depth, at the cost of some loss of quality. The idea is based on two well-known methods: Gibbs sampling and Rao-Blackwellisation. We provide some simulation-based evaluation of the quality of the RGBS with respect to run time and sequence length.
first_indexed 2025-11-14T09:10:58Z
format Conference Paper
id curtin-20.500.11937-42222
institution Curtin University Malaysia
institution_category Local University
last_indexed 2025-11-14T09:10:58Z
publishDate 2009
publisher unknown
recordtype eprints
repository_type Digital Repository
spelling curtin-20.500.11937-422222022-12-09T07:12:34Z MCMC for Hierarchical Semi-Markov Conditional Random fields Truyen, Tran Phung, Dinh Venkatesh, Svetha Bui, Hung H. Li Deng Dong Yu Geoff Hinton Deep architecture such as hierarchical semi-Markov models is an important class of models for nested sequential data. Current exact inference schemes either cost cubic time in sequence length, or exponential time in model depth. These costs are prohibitive for large-scale problems with arbitrary length and depth. In this contribution, we propose a new approximation technique that may have the potential to achieve sub-cubic time complexity in length and linear time depth, at the cost of some loss of quality. The idea is based on two well-known methods: Gibbs sampling and Rao-Blackwellisation. We provide some simulation-based evaluation of the quality of the RGBS with respect to run time and sequence length. 2009 Conference Paper http://hdl.handle.net/20.500.11937/42222 unknown fulltext
spellingShingle Truyen, Tran
Phung, Dinh
Venkatesh, Svetha
Bui, Hung H.
MCMC for Hierarchical Semi-Markov Conditional Random fields
title MCMC for Hierarchical Semi-Markov Conditional Random fields
title_full MCMC for Hierarchical Semi-Markov Conditional Random fields
title_fullStr MCMC for Hierarchical Semi-Markov Conditional Random fields
title_full_unstemmed MCMC for Hierarchical Semi-Markov Conditional Random fields
title_short MCMC for Hierarchical Semi-Markov Conditional Random fields
title_sort mcmc for hierarchical semi-markov conditional random fields
url http://hdl.handle.net/20.500.11937/42222