Closed-form Marginal Likelihood in Gamma-Poisson Matrix Factorization - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2018

Closed-form Marginal Likelihood in Gamma-Poisson Matrix Factorization

Louis Filstroff
  • Fonction : Auteur
  • PersonId : 1272587
  • IdRef : 242769659
Alberto Lumbreras

Résumé

We present novel understandings of the Gamma-Poisson (GaP) model, a probabilistic matrix fac-torization model for count data. We show that GaP can be rewritten free of the score/activation matrix. This gives us new insights about the estimation of the topic/dictionary matrix by maximum marginal likelihood estimation. In particular , this explains the robustness of this estima-tor to over-specified values of the factorization rank, especially its ability to automatically prune irrelevant dictionary columns, as empirically observed in previous work. The marginalization of the activation matrix leads in turn to a new Monte Carlo Expectation-Maximization algorithm with favorable properties.
Fichier principal
Vignette du fichier
1801.01799-15.pdf (557.02 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)
Loading...

Dates et versions

hal-02376811 , version 1 (22-11-2019)

Identifiants

  • HAL Id : hal-02376811 , version 1

Citer

Louis Filstroff, Alberto Lumbreras, Cédric Févotte. Closed-form Marginal Likelihood in Gamma-Poisson Matrix Factorization. Proc. International Conference on Machine Learning (ICML), 2018, Stockholm, Sweden. ⟨hal-02376811⟩
57 Consultations
41 Téléchargements

Partager

Gmail Facebook X LinkedIn More