Towards a filter-based EM-algorithm for parameter estimation of Littlewood's software reliability model

James Ledoux (1, 2)

1 IRMAR - Institut de Recherche Mathématique de Rennes
2 INSA Rennes - Institut National des Sciences Appliquées - Rennes

In this paper, we deal with a continuous-time software reliability model designed by Littlewood. This model may be thought of as a partially observed Markov process. The EM-algorithm is a standard way to estimate the parameters of processes with missing data. The E-step requires the computation of basic statistics related to observed/hidden processes. In this paper, we provide finite-dimensional non-linear filters for these statistics using the innovations method. This allows to plan the use of the filter-based EM-algorithm developed by Elliott.

Mots clés: Filtering; Hidden Markov process; Point process; Innovations method

Domaines: Méthodologie [stat.ME]; Performance et fiabilité [cs.PF]; Probabilités [math.PR]

James Ledoux. Towards a filter-based EM-algorithm for parameter estimation of Littlewood's software reliability model. Applied Stochastic Models and Data Analysis, May 2005, Brest, France. pp.1224-1231. ⟨hal-00853687⟩