GRU-ODE-Bayes: Continuous modeling of sporadically-observed time series. de Brouwer, E., Simm, J., Arany, A., & Moreau, Y. In Wallach, H, Larochelle, H, Beygelzimer, A, d\$\textbackslashbackslash\$textquotesingle Alché-Buc, F, Fox, E, & Garnett, R, editors, Advances in Neural Information Processing Systems, volume 32, pages 7377–7388. Curran Associates, Inc., 2019. ISSN: 10495258 _eprint: 1905.12374![GRU-ODE-Bayes: Continuous modeling of sporadically-observed time series [link]](https://bibbase.org/img/filetypes/link.svg "link")
Paper abstract bibtex Modeling real-world multidimensional time series can be particularly challenging when these are sporadically observed (i.e., sampling is irregular both in time and across dimensions)-such as in the case of clinical patient data. To address these challenges, we propose (1) a continuous-time version of the Gated Recurrent Unit, building upon the recent Neural Ordinary Differential Equations (Chen et al., 2018), and (2) a Bayesian update network that processes the sporadic observations. We bring these two ideas together in our GRU-ODE-Bayes method. We then demonstrate that the proposed method encodes a continuity prior for the latent process and that it can exactly represent the Fokker-Planck dynamics of complex processes driven by a multidimensional stochastic differential equation. Additionally, empirical evaluation shows that our method outperforms the state of the art on both synthetic data and real-world data with applications in healthcare and climate forecast. What is more, the continuity prior is shown to be well suited for low number of samples settings.
@incollection{de_brouwer_gru-ode-bayes_2019,
title = {{GRU}-{ODE}-{Bayes}: {Continuous} modeling of sporadically-observed time series},
volume = {32},
url = {http://papers.nips.cc/paper/8957-gru-ode-bayes-continuous-modeling-of-sporadically-observed-time-series.pdf http://papers.nips.cc/paper/8957-gru-ode-bayes-continuous-modeling-of-sporadically-observed-time-series https://papers.nips.cc/paper/8957-gru-ode-b},
abstract = {Modeling real-world multidimensional time series can be particularly challenging when these are sporadically observed (i.e., sampling is irregular both in time and across dimensions)-such as in the case of clinical patient data. To address these challenges, we propose (1) a continuous-time version of the Gated Recurrent Unit, building upon the recent Neural Ordinary Differential Equations (Chen et al., 2018), and (2) a Bayesian update network that processes the sporadic observations. We bring these two ideas together in our GRU-ODE-Bayes method. We then demonstrate that the proposed method encodes a continuity prior for the latent process and that it can exactly represent the Fokker-Planck dynamics of complex processes driven by a multidimensional stochastic differential equation. Additionally, empirical evaluation shows that our method outperforms the state of the art on both synthetic data and real-world data with applications in healthcare and climate forecast. What is more, the continuity prior is shown to be well suited for low number of samples settings.},
booktitle = {Advances in {Neural} {Information} {Processing} {Systems}},
publisher = {Curran Associates, Inc.},
author = {de Brouwer, Edward and Simm, Jaak and Arany, Adam and Moreau, Yves},
editor = {Wallach, H and Larochelle, H and Beygelzimer, A and d\${\textbackslash}backslash\$textquotesingle Alché-Buc, F and Fox, E and Garnett, R},
year = {2019},
note = {ISSN: 10495258
\_eprint: 1905.12374},
keywords = {\#nosource, CD GAN Review, CD GAN sort, F-Read, Review/Sporadic time-series, T-sporadic, roam},
pages = {7377--7388},
}
Downloads: 0
{"_id":"ZsFDNf8yko5sSxQtf","bibbaseid":"debrouwer-simm-arany-moreau-gruodebayescontinuousmodelingofsporadicallyobservedtimeseries-2019","authorIDs":[],"author_short":["de Brouwer, E.","Simm, J.","Arany, A.","Moreau, Y."],"bibdata":{"bibtype":"incollection","type":"incollection","title":"GRU-ODE-Bayes: Continuous modeling of sporadically-observed time series","volume":"32","url":"http://papers.nips.cc/paper/8957-gru-ode-bayes-continuous-modeling-of-sporadically-observed-time-series.pdf http://papers.nips.cc/paper/8957-gru-ode-bayes-continuous-modeling-of-sporadically-observed-time-series https://papers.nips.cc/paper/8957-gru-ode-b","abstract":"Modeling real-world multidimensional time series can be particularly challenging when these are sporadically observed (i.e., sampling is irregular both in time and across dimensions)-such as in the case of clinical patient data. To address these challenges, we propose (1) a continuous-time version of the Gated Recurrent Unit, building upon the recent Neural Ordinary Differential Equations (Chen et al., 2018), and (2) a Bayesian update network that processes the sporadic observations. We bring these two ideas together in our GRU-ODE-Bayes method. We then demonstrate that the proposed method encodes a continuity prior for the latent process and that it can exactly represent the Fokker-Planck dynamics of complex processes driven by a multidimensional stochastic differential equation. Additionally, empirical evaluation shows that our method outperforms the state of the art on both synthetic data and real-world data with applications in healthcare and climate forecast. What is more, the continuity prior is shown to be well suited for low number of samples settings.","booktitle":"Advances in Neural Information Processing Systems","publisher":"Curran Associates, Inc.","author":[{"propositions":["de"],"lastnames":["Brouwer"],"firstnames":["Edward"],"suffixes":[]},{"propositions":[],"lastnames":["Simm"],"firstnames":["Jaak"],"suffixes":[]},{"propositions":[],"lastnames":["Arany"],"firstnames":["Adam"],"suffixes":[]},{"propositions":[],"lastnames":["Moreau"],"firstnames":["Yves"],"suffixes":[]}],"editor":[{"propositions":[],"lastnames":["Wallach"],"firstnames":["H"],"suffixes":[]},{"propositions":[],"lastnames":["Larochelle"],"firstnames":["H"],"suffixes":[]},{"propositions":[],"lastnames":["Beygelzimer"],"firstnames":["A"],"suffixes":[]},{"propositions":["d\\$\\textbackslashbackslash\\$textquotesingle"],"lastnames":["Alché-Buc"],"firstnames":["F"],"suffixes":[]},{"propositions":[],"lastnames":["Fox"],"firstnames":["E"],"suffixes":[]},{"propositions":[],"lastnames":["Garnett"],"firstnames":["R"],"suffixes":[]}],"year":"2019","note":"ISSN: 10495258 _eprint: 1905.12374","keywords":"#nosource, CD GAN Review, CD GAN sort, F-Read, Review/Sporadic time-series, T-sporadic, roam","pages":"7377–7388","bibtex":"@incollection{de_brouwer_gru-ode-bayes_2019,\n\ttitle = {{GRU}-{ODE}-{Bayes}: {Continuous} modeling of sporadically-observed time series},\n\tvolume = {32},\n\turl = {http://papers.nips.cc/paper/8957-gru-ode-bayes-continuous-modeling-of-sporadically-observed-time-series.pdf http://papers.nips.cc/paper/8957-gru-ode-bayes-continuous-modeling-of-sporadically-observed-time-series https://papers.nips.cc/paper/8957-gru-ode-b},\n\tabstract = {Modeling real-world multidimensional time series can be particularly challenging when these are sporadically observed (i.e., sampling is irregular both in time and across dimensions)-such as in the case of clinical patient data. To address these challenges, we propose (1) a continuous-time version of the Gated Recurrent Unit, building upon the recent Neural Ordinary Differential Equations (Chen et al., 2018), and (2) a Bayesian update network that processes the sporadic observations. We bring these two ideas together in our GRU-ODE-Bayes method. We then demonstrate that the proposed method encodes a continuity prior for the latent process and that it can exactly represent the Fokker-Planck dynamics of complex processes driven by a multidimensional stochastic differential equation. Additionally, empirical evaluation shows that our method outperforms the state of the art on both synthetic data and real-world data with applications in healthcare and climate forecast. What is more, the continuity prior is shown to be well suited for low number of samples settings.},\n\tbooktitle = {Advances in {Neural} {Information} {Processing} {Systems}},\n\tpublisher = {Curran Associates, Inc.},\n\tauthor = {de Brouwer, Edward and Simm, Jaak and Arany, Adam and Moreau, Yves},\n\teditor = {Wallach, H and Larochelle, H and Beygelzimer, A and d\\${\\textbackslash}backslash\\$textquotesingle Alché-Buc, F and Fox, E and Garnett, R},\n\tyear = {2019},\n\tnote = {ISSN: 10495258\n\\_eprint: 1905.12374},\n\tkeywords = {\\#nosource, CD GAN Review, CD GAN sort, F-Read, Review/Sporadic time-series, T-sporadic, roam},\n\tpages = {7377--7388},\n}\n\n","author_short":["de Brouwer, E.","Simm, J.","Arany, A.","Moreau, Y."],"editor_short":["Wallach, H","Larochelle, H","Beygelzimer, A","d\\$\\textbackslashbackslash\\$textquotesingle Alché-Buc, F","Fox, E","Garnett, R"],"key":"de_brouwer_gru-ode-bayes_2019","id":"de_brouwer_gru-ode-bayes_2019","bibbaseid":"debrouwer-simm-arany-moreau-gruodebayescontinuousmodelingofsporadicallyobservedtimeseries-2019","role":"author","urls":{"Paper":"http://papers.nips.cc/paper/8957-gru-ode-bayes-continuous-modeling-of-sporadically-observed-time-series.pdf http://papers.nips.cc/paper/8957-gru-ode-bayes-continuous-modeling-of-sporadically-observed-time-series https://papers.nips.cc/paper/8957-gru-ode-b"},"keyword":["#nosource","CD GAN Review","CD GAN sort","F-Read","Review/Sporadic time-series","T-sporadic","roam"],"downloads":0},"bibtype":"incollection","biburl":"https://api.zotero.org/users/3522498/collections/DIPSR9D9/items?key=GIOIfx1olqrQ1vxbE1TrXAIi&format=bibtex&limit=100","creationDate":"2021-03-03T03:31:06.370Z","downloads":0,"keywords":["#nosource","cd gan review","cd gan sort","f-read","review/sporadic time-series","t-sporadic","roam"],"search_terms":["gru","ode","bayes","continuous","modeling","sporadically","observed","time","series","de brouwer","simm","arany","moreau"],"title":"GRU-ODE-Bayes: Continuous modeling of sporadically-observed time series","year":2019,"dataSources":["zeSfpfAFdsRF2mNyj"]}