Bayesian inference for linear dynamic models with Dirichlet process mixtures
Using Kalman techniques, it is possible to perform optimal estimation in linear Gaussian state-space models. Here, we address the case where the noise probability density functions are of unknown functional form. A flexible Bayesian nonparametric noise model based on Dirichlet process mixtures is in...
| Main Authors: | , , , , |
|---|---|
| Format: | Journal article |
| Language: | English |
| Published: |
2008
|
| _version_ | 1797091160608473088 |
|---|---|
| author | Caron, F Davy, M Doucet, A Duflos, E Vanheeghe, P |
| author_facet | Caron, F Davy, M Doucet, A Duflos, E Vanheeghe, P |
| author_sort | Caron, F |
| collection | OXFORD |
| description | Using Kalman techniques, it is possible to perform optimal estimation in linear Gaussian state-space models. Here, we address the case where the noise probability density functions are of unknown functional form. A flexible Bayesian nonparametric noise model based on Dirichlet process mixtures is introduced. Efficient Markov chain Monte Carlo and sequential Monte Carlo methods are then developed to perform optimal batch and sequential estimation in such contexts. The algorithms are applied to blind deconvolution and change point detection. Experimental results on synthetic and real data demonstrate the efficiency of this approach in various contexts. © 2007 IEEE. |
| first_indexed | 2024-03-07T03:29:03Z |
| format | Journal article |
| id | oxford-uuid:ba0d93d1-858c-444e-8183-f694023cff91 |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-07T03:29:03Z |
| publishDate | 2008 |
| record_format | dspace |
| spelling | oxford-uuid:ba0d93d1-858c-444e-8183-f694023cff912022-03-27T05:07:21ZBayesian inference for linear dynamic models with Dirichlet process mixturesJournal articlehttp://purl.org/coar/resource_type/c_dcae04bcuuid:ba0d93d1-858c-444e-8183-f694023cff91EnglishSymplectic Elements at Oxford2008Caron, FDavy, MDoucet, ADuflos, EVanheeghe, PUsing Kalman techniques, it is possible to perform optimal estimation in linear Gaussian state-space models. Here, we address the case where the noise probability density functions are of unknown functional form. A flexible Bayesian nonparametric noise model based on Dirichlet process mixtures is introduced. Efficient Markov chain Monte Carlo and sequential Monte Carlo methods are then developed to perform optimal batch and sequential estimation in such contexts. The algorithms are applied to blind deconvolution and change point detection. Experimental results on synthetic and real data demonstrate the efficiency of this approach in various contexts. © 2007 IEEE. |
| spellingShingle | Caron, F Davy, M Doucet, A Duflos, E Vanheeghe, P Bayesian inference for linear dynamic models with Dirichlet process mixtures |
| title | Bayesian inference for linear dynamic models with Dirichlet process mixtures |
| title_full | Bayesian inference for linear dynamic models with Dirichlet process mixtures |
| title_fullStr | Bayesian inference for linear dynamic models with Dirichlet process mixtures |
| title_full_unstemmed | Bayesian inference for linear dynamic models with Dirichlet process mixtures |
| title_short | Bayesian inference for linear dynamic models with Dirichlet process mixtures |
| title_sort | bayesian inference for linear dynamic models with dirichlet process mixtures |
| work_keys_str_mv | AT caronf bayesianinferenceforlineardynamicmodelswithdirichletprocessmixtures AT davym bayesianinferenceforlineardynamicmodelswithdirichletprocessmixtures AT douceta bayesianinferenceforlineardynamicmodelswithdirichletprocessmixtures AT duflose bayesianinferenceforlineardynamicmodelswithdirichletprocessmixtures AT vanheeghep bayesianinferenceforlineardynamicmodelswithdirichletprocessmixtures |