Joint modeling of multiple time series via the beta process with application to motion capture segmentation

DOI10.1214/14-AOAS742zbMATH Open1303.62048arXiv1308.4747MaRDI QIDQ483980FDOQ483980

Authors: Emily B. Fox, Michael C. Hughes, Erik B. Sudderth, Michael Jordan 

Publication date: 17 December 2014

Published in: The Annals of Applied Statistics (Search for Journal in Brave)

Abstract: We propose a Bayesian nonparametric approach to the problem of jointly modeling multiple related time series. Our model discovers a latent set of dynamical behaviors shared among the sequences, and segments each time series into regions defined by a subset of these behaviors. Using a beta process prior, the size of the behavior set and the sharing pattern are both inferred from data. We develop Markov chain Monte Carlo (MCMC) methods based on the Indian buffet process representation of the predictive distribution of the beta process. Our MCMC inference algorithm efficiently adds and removes behaviors via novel split-merge moves as well as data-driven birth and death proposals, avoiding the need to consider a truncated model. We demonstrate promising results on unsupervised segmentation of human motion capture data.

Full work available at URL: https://arxiv.org/abs/1308.4747

Recommendations

• Beta-product dependent Pitman-Yor processes for Bayesian inference
• Modeling and recognizing human trajectories with beta process hidden Markov models
• A factor model approach for the joint segmentation with between-series correlation
• A Bayesian non-parametric dynamic AR model for multiple time series analysis
• Joint segmentation of multivariate Gaussian processes using mixed linear models

zbMATH Keywords

Bayesian nonparametricshidden Markov modelsmotion capturemultiple time seriesbeta process

Mathematics Subject Classification ID

Computational methods in Markov chains (60J22) Bayesian inference (62F15) Nonparametric estimation (62G05) Markov processes: estimation; hidden Markov models (62M05) Time series, auto-correlation, regression, etc. in statistics (GARCH) (62M10) Random measures (60G57)

Cites Work

• Reversible jump Markov chain Monte Carlo computation and Bayesian model determination
• Markov chains for exploring posterior distributions. (With discussion)
• Order-Based Dependent Dirichlet Processes
• Completely random measures
• Mixed Hidden Markov Models
• A sticky HDP-HMM with application to speaker diarization
• Bayesian forecasting and dynamic models.
• Title not available (Why is that?)
• Hierarchical Dirichlet Processes
• Splitting and merging components of a nonconjugate Dirichlet process mixture model
• Nonparametric Bayes estimators based on beta processes in models for life history data
• Title not available (Why is that?)
• Nonparametric Bayes local partition models for random effects
• Factorial hidden Markov models
• Title not available (Why is that?)
• Music Analysis Using Hidden Markov Mixture Models
• Multivariate kernel partition process mixtures
• State space modeling of multiple time series
• Miscellanea. Peskun's theorem and a modified discrete-state Gibbs sampler
• Joint modeling of multiple time series via the beta process with application to motion capture segmentation
• Title not available (Why is that?)

Cited In (13)

• Model structure selection for switched NARX system identification: a randomized approach
• Generalized species sampling priors with latent beta reinforcements
• A grouped beta process model for multivariate resting‐state EEG microstate analysis on twins
• Bayesian sparse vector autoregressive switching models with application to human gesture phase segmentation
• A nonparametrically corrected likelihood for Bayesian spectral analysis of multivariate time series
• Statistical analysis of multi-day solar irradiance using a threshold time series model
• Beta-product dependent Pitman-Yor processes for Bayesian inference
• Modeling and recognizing human trajectories with beta process hidden Markov models
• Title not available (Why is that?)
• Accelerated parallel non-conjugate sampling for Bayesian non-parametric models
• Estimation of jump Box-Jenkins models
• Joint modeling of multiple time series via the beta process with application to motion capture segmentation
• Title not available (Why is that?)

Uses Software

• hmm
• CMU Motion Capture

This page was built for publication: Joint modeling of multiple time series via the beta process with application to motion capture segmentation

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q483980)