Title: A purely Bayesian approach for proportional visual data modelling

Authors: Sami Bourouis; Yacine Laalaoui; Nizar Bouguila

Addresses: Department of Information Technology, College of Computers and Information Technology, Taif University, Taif, KSA ' Department of Information Technology, College of Computers and Information Technology, Taif University, Taif, KSA ' The Concordia Institute for Information Systems Engineering (CIISE), Concordia University, Montreal, QC, H3G 1T7, Canada

Abstract: In this paper, we focus on constructing new flexible and powerful parametric framework for proportional visual data modelling. In particular, we propose a Bayesian density estimation method based upon mixtures of scaled Dirichlet distributions. The consideration of Bayesian learning is interesting in several aspects. It allows simultaneous parameters estimation and model selection, it permits also taking uncertainty into account by introducing prior information about the parameters and it allows overcoming learning problems related to over- or under-fitting. In this work, three key issues related to the Bayesian mixture learning are addressed which are the choice of prior distributions, the estimation of the parameters, and the selection of the number of components. Moreover, a principled Metropolis-within-Gibbs sampler algorithm for scaled Dirichlet mixtures is developed. Finally, the proposed Bayesian framework is tested via two challenging real-life applications namely scene reconstruction and face age estimation from images. The obtained results show the merits of our approach.

Keywords: mixture models; scaled Dirichlet; Bayesian inference; Gibbs sampling; Metropolis-Hastings; scene reconstruction; face age estimation.

DOI: 10.1504/IJIEI.2018.094513

International Journal of Intelligent Engineering Informatics, 2018 Vol.6 No.5, pp.491 - 508

Received: 04 Apr 2018
Accepted: 11 Apr 2018

Published online: 04 Sep 2018 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article