Title: Some remarks about maximization of the likelihood function
Authors: Adelchi Azzalini - Università di Padova (Italy) [presenting]
Abstract: In most practical cases, maximum likelihood estimation is performed via some numerical method. Within this frame, there exists a range of alternatives in current use, and the question of choosing the more convenient option for a given problem does not usually have a clear-cut answer. We discuss some issues and points to consider in this context, with special emphasis on the contrast between EM-type algorithms and those based on direct numerical maximization.