Probably Approximately Correct Learning of Regulatory Networks from Time-Series Data

Abstract : Automating the process of model building from experimental data is a very desirable goal to palliate the lack of modellers for many applications. However, despite the spectacular progress of machine learning techniques in data analytics, classification, clustering and prediction making, learning dynamical models from data time-series is still challenging. In this paper we investigate the use of the Probably Approximately Correct (PAC) learning framework of Leslie Valiant as a method for the automated discovery of influence models of biochemical processes from Boolean and stochastic traces. We show that Thomas' Boolean influence systems can be naturally represented by k-CNF formulae, and learned from time-series data with a number of Boolean activation samples per species quasi-linear in the precision of the learned model, and that positive Boolean influence systems can be represented by monotone DNF formulae and learned actively with both activation samples and oracle calls. We consider Boolean traces and Boolean abstractions of stochastic simulation traces, and study the space-time tradeoff there is between the diversity of initial states and the length of the time horizon, and its impact on the error bounds provided by the PAC learning algorithms. We evaluate the performance of this approach on a model of T-lymphocyte differentiation, with and without prior knowledge, and discuss its merits as well as its limitations with respect to realistic experiments.
Type de document :
Communication dans un congrès
J. Feret and H. Koeppl. CMSB'17 - 15th International Conference on Computational Methods for Systems Biology, Sep 2017, Darmstadt, Germany. Lecture Notes in Computer Science (10545), pp.74-90
Liste complète des métadonnées

Littérature citée [25 références]  Voir  Masquer  Télécharger

https://hal.inria.fr/hal-01519826
Contributeur : François Fages <>
Soumis le : mardi 4 juillet 2017 - 14:35:41
Dernière modification le : lundi 24 septembre 2018 - 14:00:04
Document(s) archivé(s) le : jeudi 14 décembre 2017 - 23:41:33

Fichier

CFS17cmsb.pdf
Fichiers produits par l'(les) auteur(s)

Identifiants

  • HAL Id : hal-01519826, version 2

Citation

Arthur Carcano, François Fages, Sylvain Soliman. Probably Approximately Correct Learning of Regulatory Networks from Time-Series Data. J. Feret and H. Koeppl. CMSB'17 - 15th International Conference on Computational Methods for Systems Biology, Sep 2017, Darmstadt, Germany. Lecture Notes in Computer Science (10545), pp.74-90. 〈hal-01519826v2〉

Partager

Métriques

Consultations de la notice

786

Téléchargements de fichiers

365