Hidden Conditional Random Fields

Quattoni, Ariadna; Wang, S.; Morency, Louis–Philippe; Collins, Michael; Darrell, Trevor

doi:10.1109/tpami.2007.1124

Cited by 461 publications

(403 citation statements)

References 13 publications

Supporting

Mentioning

397

Contrasting

Unclassified

Order By: Relevance

“…HCRFs -discriminative undirected models that contain hidden states-were first presented in [10] and used to capture temporal dependencies across frames and recognize different gesture classes. They did so successfully by learning a state distribution among the different gesture classes in a discriminative manner, allowing them to not only uncover the distinctive configurations that uniquely identify each class, but also to learn a shared common structure among the classes.…”

Section: Finite Hidden Conditional Random Fieldsmentioning

confidence: 99%

“…We did not use regularization for our HCRF-DPM models and all of them had their truncation level set to L = 10 and their hyperparameters to s 1 = 1000 and s 2 = 10. Finally, our finite HCRF models were trained with a maximum of 300 iterations for the gradient ascent method used [10], whereas our HCRF-DPM models were trained with a maximum of 1200 variational coordinate descent iterations and a maximum of 600 iterations of gradient descent. All IHCRF-MCMC models were trained according to the experimental protocol of [7].…”

Section: Application To the Audiovisual Analysis Of Human Behaviormentioning

confidence: 99%

See 1 more Smart Citation

Variational Hidden Conditional Random Fields with Coupled Dirichlet Process Mixtures

Bousmalis

Zafeiriou

Morency

et al. 2013

Advanced Information Systems Engineering

View full text Add to dashboard Cite

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 HCRF is an HCRF 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.

show abstract

Section: Finite Hidden Conditional Random Fieldsmentioning

confidence: 99%

Section: Application To the Audiovisual Analysis Of Human Behaviormentioning

confidence: 99%

Variational Hidden Conditional Random Fields with Coupled Dirichlet Process Mixtures

Bousmalis

Zafeiriou

Morency

et al. 2013

Advanced Information Systems Engineering

View full text Add to dashboard Cite

show abstract

“…Hidden conditional random field was first introduced by Gunawardana et al [11] for phone-conversation/speech classification and has then been applied to gesture and object recognition [24,15]. Given a sequence composed of a set of n local observations {x 1 , x 2 , x 3 , ... , x n } denoted by X, and its class labels y ∈ Y, we want to find a mapping p(y|X) between them, where y is conditioned on X.…”

Section: Hidden Conditional Random Fieldmentioning

confidence: 99%

“…To overcome this limitation, conditional random field (CRF) was recently introduced [22,20]. However, CRF cannot incorporate the need for labelling a whole sequence as an action, and also cannot capture the intermediate structures using hidden state variables [15]. To overcome these problems, hidden conditional random fields (HCRF) was proposed by [11,24,15].…”

Section: Introductionmentioning

confidence: 99%

“…However, CRF cannot incorporate the need for labelling a whole sequence as an action, and also cannot capture the intermediate structures using hidden state variables [15]. To overcome these problems, hidden conditional random fields (HCRF) was proposed by [11,24,15]. Compared to CRF, HCRF is capable of incorporating a sequence label into the optimization of observation conditional probabilities.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Action categorization with modified hidden conditional random field

Zhang

Gong

2010

Pattern Recognition

View full text Add to dashboard Cite

In this paper, we present a method for action categorization with a modified hidden conditional random field (HCRF). Specifically, effective silhouette-based action features are extracted using motion moments and spectrum of chain code. We formulate a modified HCRF (mHCRF) to have a guaranteed global optimum in the modelling of the temporal action dependencies after the HMM pathing stage. Experimental results on action categorization using this model are compared favorably against several existing modelbased methods including GMM, SVM, Logistic Regression, HMM, CRF and HCRF.

show abstract

References

2019

Opinion Analysis in Interactions

View full text Add to dashboard Cite

Hidden Conditional Random Fields

Cited by 461 publications

References 13 publications

Variational Hidden Conditional Random Fields with Coupled Dirichlet Process Mixtures

Variational Hidden Conditional Random Fields with Coupled Dirichlet Process Mixtures

Action categorization with modified hidden conditional random field

References

Contact Info

Product

Resources

About