We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ML

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo

Statistics > Machine Learning

Title: Neural graphical modelling in continuous-time: consistency guarantees and algorithms

Abstract: The discovery of structure from time series data is a key problem in fields of study working with complex systems. Most identifiability results and learning algorithms assume the underlying dynamics to be discrete in time. Comparatively few, in contrast, explicitly define dependencies in infinitesimal intervals of time, independently of the scale of observation and of the regularity of sampling. In this paper, we consider score-based structure learning for the study of dynamical systems. We prove that for vector fields parameterized in a large class of neural networks, least squares optimization with adaptive regularization schemes consistently recovers directed graphs of local independencies in systems of stochastic differential equations. Using this insight, we propose a score-based learning algorithm based on penalized Neural Ordinary Differential Equations (modelling the mean process) that we show to be applicable to the general setting of irregularly-sampled multivariate time series and to outperform the state of the art across a range of dynamical systems.
Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Dynamical Systems (math.DS)
Cite as: arXiv:2105.02522 [stat.ML]
  (or arXiv:2105.02522v3 [stat.ML] for this version)

Submission history

From: Alexis Bellot [view email]
[v1] Thu, 6 May 2021 08:48:02 GMT (809kb,D)
[v2] Tue, 28 Sep 2021 21:33:12 GMT (1492kb,D)
[v3] Thu, 3 Feb 2022 15:19:54 GMT (1575kb,D)

Link back to: arXiv, form interface, contact.