Current browse context:
stat.CO
Change to browse by:
References & Citations
Statistics > Methodology
Title: Latent variable model selection for Gaussian conditional random fields
(Submitted on 20 Dec 2015 (v1), last revised 4 Mar 2017 (this version, v3))
Abstract: We consider the problem of learning a conditional Gaussian graphical model in the presence of latent variables. Building on recent advances in this field, we suggest a method that decomposes the parameters of a conditional Markov random field into the sum of a sparse and a low-rank matrix. We derive convergence bounds for this estimator and show that it is well-behaved in the high-dimensional regime as well as "sparsistent" (i.e. capable of recovering the graph structure). We then show how proximal gradient algorithms and semi-definite programming techniques can be employed to fit the model to thousands of variables. Through extensive simulations, we illustrate the conditions required for identifiability and show that there is a wide range of situations in which this model performs significantly better than its counterparts, for example, by accommodating more latent variables. Finally, the suggested method is applied to two datasets comprising individual level data on genetic variants and metabolites levels. We show our results replicate better than alternative approaches and show enriched biological signal.
Submission history
From: Benjamin Frot [view email][v1] Sun, 20 Dec 2015 17:44:49 GMT (289kb)
[v2] Thu, 17 Mar 2016 00:24:33 GMT (3911kb)
[v3] Sat, 4 Mar 2017 21:49:35 GMT (1091kb)
Link back to: arXiv, form interface, contact.