On recursive estimation in incomplete data models

We consider a new recursive algorithm for parameter estimation from an inde pendent incomplete data sequence. The algorithm can be viewed as a recursiv e version of the well-known EM algorithm, augmented with a Monte-Carlo step which restores the missing data. Based on recent results on stochastic alg orithms, we give conditions for the a.s. convergence of the algorithm. More over, asymptotical variance of this estimator is reduced by a simple averag ing. Application to finite mixtures is given with a simulation experiment.