AUEB STATS SEMINARS 21/3/2018: Learning Markov Equivalence Classes of Directed Acyclic Graphs: an Objective Bayes Approach by Prof. Guido Consonni
Mon 12 Mar 2018 - 12:31
ΚΥΚΛΟΣ ΣΕΜΙΝΑΡΙΩΝ ΣΤΑΤΙΣΤΙΚΗΣ ΜΑΡΤΙΟΣ 2018
Guido Consonni
Universita' Cattolica del Sacro Cuore, Milan, Italy
Learning Markov Equivalence Classes of Directed Acyclic Graphs: an Objective Bayes Approach
ΤΕΤΑΡΤΗ 21/3/2018
13:30
ΑΙΘΟΥΣΑ T201, 2ος ΟΡΟΦΟΣ,
ΝΕΟ ΚΤΙΡΙΟ Ο.Π.Α. (ΤΡΟΙΑΣ 2)
ΠΕΡΙΛΗΨΗ
A Markov equivalence class contains all the Directed Acyclic Graphs (DAGs) encoding the same conditional independencies, and is represented by a Completed Partially Directed DAG (CPDAG), also named Essential Graph (EG).
We approach the problem of model selection among noncausal sparse Gaussian DAGs by directly scoring EGs, using an objective Bayes method. Specifically, we construct objective priors for model selection based on the Fractional Bayes Factor, leading to a closed form expression for the marginal likelihood of an EG.
Next we propose an MCMC strategy to explore the space of EGs, possibly accounting for sparsity constraints, and illustrate the performance of our method on simulation studies, as well as on a real dataset.
Our method is fully Bayesian and thus provides a coherent quantification of inferential uncertainty, requires minimal prior specification, and shows to be competitive in learning the structure of the data-generating EG when compared to alternative state-of-the-art algorithms.
- AUEB STATS SEMINARS 12/12/2018: A new approach to Bayesian image analysis by John Kornak
- AUEB STATS SEMINARS 10/1/2019: Group invariance for objective prior elicitation in testing problems by Anabel Forte
- AUEB STATS SEMINARS 16/10/2018: Phase II trials in oncology – is Simon’s optimal design the optimum? by Prof. Meinhard Kieser
- AUEB STATS SEMINARS 10/5/2018: Interlocking directorates in Irish companies using a latent space model for bipartite networks by Prof. Nial Friel
- AUEB STATS SEMINARS 30/3/2018: Estimating metabolite networks using a multi-step network approach that integrates information from lower leveled omic graphical structures by Georgios Bartzis
Permissions in this forum:
You cannot reply to topics in this forum