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


Current browse context:


Change to browse by:

References & Citations


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

Statistics > Machine Learning

Title: Towards a mathematical theory of trajectory inference

Abstract: We devise a theoretical framework and a numerical method to infer trajectories of a stochastic process from snapshots of its temporal marginals. This problem arises in the analysis of single cell RNA-sequencing data, which provide high dimensional measurements of cell states but cannot track the trajectories of the cells over time. We prove that for a class of stochastic processes it is possible to recover the ground truth trajectories from limited samples of the temporal marginals at each time-point, and provide an efficient algorithm to do so in practice. The method we develop, Global Waddington-OT (gWOT), boils down to a smooth convex optimization problem posed globally over all time-points involving entropy-regularized optimal transport. We demonstrate that this problem can be solved efficiently in practice and yields good reconstructions, as we show on several synthetic and real datasets.
Comments: The first two authors contributed equally to this work; 62 pages
Subjects: Machine Learning (stat.ML); Machine Learning (cs.LG); Optimization and Control (math.OC); Probability (math.PR); Statistics Theory (math.ST)
MSC classes: 62M20 (Primary) 62G99, 49M29, 92C15 (Secondary)
Cite as: arXiv:2102.09204 [stat.ML]
  (or arXiv:2102.09204v1 [stat.ML] for this version)

Submission history

From: Hugo Lavenant [view email]
[v1] Thu, 18 Feb 2021 07:58:47 GMT (1854kb,D)

Link back to: arXiv, form interface, contact.