A comparison of computational approaches for maximum likelihood estimation of the Dirichlet parameters on high-dimensional data

• Marco Giordan ^[1] ; Ron Wehrens ^[1]
  1. [1] IASMA Research and Innovation Centre
• Localización: Sort: Statistics and Operations Research Transactions, ISSN 1696-2281, Vol. 39, Nº. 1, 2015, págs. 109-126
• Idioma: inglés
• Enlaces
  • Texto completo (pdf)
• Resumen

  • Likelihood estimates of the Dirichlet distribution parameters can be obtained only through numer- ical algorithms. Such algorithms can provide estimates outside the correct range for the parame- ters and/or can require a large amount of iterations to reach convergence. These problems can be aggravated if good starting values are not provided. In this paper we discuss several approaches that can partially avoid these problems providing a good trade off between efficiency and stability.

    The performances of these approaches are compared on high-dimensional real and simulated data.