Titre :
"Composite
likelihood methods in Markov chain models
."
Cristiano Varin
Department of Statistics
University Ca' Foscari - Venice
Abstract:
In many spatial and spatial-temporal models, and more
generally in models with complex dependencies, it may be too difficult
to carry out full maximum likelihood analysis. Possible remedies
includes the use of the composite likelihood. This term indicates a
rich class of pseudolikelihoods constructed by composing
likelihood-type objects. In this talk, I will discuss the efficiency
properties of different type of composite likelihoods for general
Markov chain models. The project is motivated by the desire to
understand the precise behaviour of composite likelihood methods in a
simple but not trivial setting where this can be analysed without the
need of simulations. This is a joint project with Prof. Nils Lid Hjort,
University of Oslo (Norway).