Variational Infinite Hidden Conditional Random Fields. Bousmalis, K., Zafeiriou, S., Morency, L. P., Pantic, M., & Ghahramani, Z. IEEE Transactions on Pattern Analysis and Machine Intelligence, 37(9):1917--1929, September, 2015. 00000doi abstract bibtex Hidden conditional random fields (HCRFs) are discriminative latent variable models which have been shown to successfully learn the hidden structure of a given classification problem. An Infinite hidden conditional random field is a hidden conditional random field with a countably infinite number of hidden states, which rids us not only of the necessity to specify a priori a fixed number of hidden states available but also of the problem of overfitting. Markov chain Monte Carlo (MCMC) sampling algorithms are often employed for inference in such models. However, convergence of such algorithms is rather difficult to verify, and as the complexity of the task at hand increases the computational cost of such algorithms often becomes prohibitive. These limitations can be overcome by variational techniques. In this paper, we present a generalized framework for infinite HCRF models, and a novel variational inference approach on a model based on coupled Dirichlet Process Mixtures, the HCRF-DPM. We show that the variational HCRF-DPM is able to converge to a correct number of represented hidden states, and performs as well as the best parametric HCRFs—chosen via cross-validation—for the difficult tasks of recognizing instances of agreement, disagreement, and pain in audiovisual sequences.
@article{bousmalis_variational_2015,
title = {Variational {Infinite} {Hidden} {Conditional} {Random} {Fields}},
volume = {37},
issn = {0162-8828},
doi = {10.1109/TPAMI.2014.2388228},
abstract = {Hidden conditional random fields (HCRFs) are discriminative latent variable models which have been shown to successfully learn the hidden structure of a given classification problem. An Infinite hidden conditional random field is a hidden conditional random field with a countably infinite number of hidden states, which rids us not only of the necessity to specify a priori a fixed number of hidden states available but also of the problem of overfitting. Markov chain Monte Carlo (MCMC) sampling algorithms are often employed for inference in such models. However, convergence of such algorithms is rather difficult to verify, and as the complexity of the task at hand increases the computational cost of such algorithms often becomes prohibitive. These limitations can be overcome by variational techniques. In this paper, we present a generalized framework for infinite HCRF models, and a novel variational inference approach on a model based on coupled Dirichlet Process Mixtures, the HCRF-DPM. We show that the variational HCRF-DPM is able to converge to a correct number of represented hidden states, and performs as well as the best parametric HCRFs—chosen via cross-validation—for the difficult tasks of recognizing instances of agreement, disagreement, and pain in audiovisual sequences.},
number = {9},
journal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},
author = {Bousmalis, K. and Zafeiriou, S. and Morency, L. P. and Pantic, M. and Ghahramani, Z.},
month = sep,
year = {2015},
note = {00000},
keywords = {Analytical models, Computational modeling, Discriminative models, Hidden Markov models, Inference algorithms, Joints, Nonparametric models, Random variables, convergence, dirichlet processes, hidden conditional random fields, variational inference},
pages = {1917--1929}
}
Downloads: 0
{"_id":"4AbbRRcA5qE7YLfXz","bibbaseid":"bousmalis-zafeiriou-morency-pantic-ghahramani-variationalinfinitehiddenconditionalrandomfields-2015","downloads":0,"creationDate":"2017-04-24T18:33:14.833Z","title":"Variational Infinite Hidden Conditional Random Fields","author_short":["Bousmalis, K.","Zafeiriou, S.","Morency, L. P.","Pantic, M.","Ghahramani, Z."],"year":2015,"bibtype":"article","biburl":"http://bibbase.org/zotero/BobWong","bibdata":{"bibtype":"article","type":"article","title":"Variational Infinite Hidden Conditional Random Fields","volume":"37","issn":"0162-8828","doi":"10.1109/TPAMI.2014.2388228","abstract":"Hidden conditional random fields (HCRFs) are discriminative latent variable models which have been shown to successfully learn the hidden structure of a given classification problem. An Infinite hidden conditional random field is a hidden conditional random field with a countably infinite number of hidden states, which rids us not only of the necessity to specify a priori a fixed number of hidden states available but also of the problem of overfitting. Markov chain Monte Carlo (MCMC) sampling algorithms are often employed for inference in such models. However, convergence of such algorithms is rather difficult to verify, and as the complexity of the task at hand increases the computational cost of such algorithms often becomes prohibitive. These limitations can be overcome by variational techniques. In this paper, we present a generalized framework for infinite HCRF models, and a novel variational inference approach on a model based on coupled Dirichlet Process Mixtures, the HCRF-DPM. We show that the variational HCRF-DPM is able to converge to a correct number of represented hidden states, and performs as well as the best parametric HCRFs—chosen via cross-validation—for the difficult tasks of recognizing instances of agreement, disagreement, and pain in audiovisual sequences.","number":"9","journal":"IEEE Transactions on Pattern Analysis and Machine Intelligence","author":[{"propositions":[],"lastnames":["Bousmalis"],"firstnames":["K."],"suffixes":[]},{"propositions":[],"lastnames":["Zafeiriou"],"firstnames":["S."],"suffixes":[]},{"propositions":[],"lastnames":["Morency"],"firstnames":["L.","P."],"suffixes":[]},{"propositions":[],"lastnames":["Pantic"],"firstnames":["M."],"suffixes":[]},{"propositions":[],"lastnames":["Ghahramani"],"firstnames":["Z."],"suffixes":[]}],"month":"September","year":"2015","note":"00000","keywords":"Analytical models, Computational modeling, Discriminative models, Hidden Markov models, Inference algorithms, Joints, Nonparametric models, Random variables, convergence, dirichlet processes, hidden conditional random fields, variational inference","pages":"1917--1929","bibtex":"@article{bousmalis_variational_2015,\n\ttitle = {Variational {Infinite} {Hidden} {Conditional} {Random} {Fields}},\n\tvolume = {37},\n\tissn = {0162-8828},\n\tdoi = {10.1109/TPAMI.2014.2388228},\n\tabstract = {Hidden conditional random fields (HCRFs) are discriminative latent variable models which have been shown to successfully learn the hidden structure of a given classification problem. An Infinite hidden conditional random field is a hidden conditional random field with a countably infinite number of hidden states, which rids us not only of the necessity to specify a priori a fixed number of hidden states available but also of the problem of overfitting. Markov chain Monte Carlo (MCMC) sampling algorithms are often employed for inference in such models. However, convergence of such algorithms is rather difficult to verify, and as the complexity of the task at hand increases the computational cost of such algorithms often becomes prohibitive. These limitations can be overcome by variational techniques. In this paper, we present a generalized framework for infinite HCRF models, and a novel variational inference approach on a model based on coupled Dirichlet Process Mixtures, the HCRF-DPM. We show that the variational HCRF-DPM is able to converge to a correct number of represented hidden states, and performs as well as the best parametric HCRFs—chosen via cross-validation—for the difficult tasks of recognizing instances of agreement, disagreement, and pain in audiovisual sequences.},\n\tnumber = {9},\n\tjournal = {IEEE Transactions on Pattern Analysis and Machine Intelligence},\n\tauthor = {Bousmalis, K. and Zafeiriou, S. and Morency, L. P. and Pantic, M. and Ghahramani, Z.},\n\tmonth = sep,\n\tyear = {2015},\n\tnote = {00000},\n\tkeywords = {Analytical models, Computational modeling, Discriminative models, Hidden Markov models, Inference algorithms, Joints, Nonparametric models, Random variables, convergence, dirichlet processes, hidden conditional random fields, variational inference},\n\tpages = {1917--1929}\n}\n\n","author_short":["Bousmalis, K.","Zafeiriou, S.","Morency, L. P.","Pantic, M.","Ghahramani, Z."],"key":"bousmalis_variational_2015","id":"bousmalis_variational_2015","bibbaseid":"bousmalis-zafeiriou-morency-pantic-ghahramani-variationalinfinitehiddenconditionalrandomfields-2015","role":"author","urls":{},"keyword":["Analytical models","Computational modeling","Discriminative models","Hidden Markov models","Inference algorithms","Joints","Nonparametric models","Random variables","convergence","dirichlet processes","hidden conditional random fields","variational inference"],"downloads":0},"search_terms":["variational","infinite","hidden","conditional","random","fields","bousmalis","zafeiriou","morency","pantic","ghahramani"],"keywords":["analytical models","computational modeling","discriminative models","hidden markov models","inference algorithms","joints","nonparametric models","random variables","convergence","dirichlet processes","hidden conditional random fields","variational inference"],"authorIDs":[],"dataSources":["PRu9F5KuG5QWYnRZL"]}