Deterministic Annealing Variant of the EM Algorithm

We present a deterministic annealing variant of the EM algorithm maximum likelihood parameter estimation problems. In ourfor approach, the EM process is reformulated as the problem of minimizing the thermodynamic free energy by using the principle of maximum entropy and statistical mechanics analogy. Unlike simulated deterministicallyannealing approaches, this minimization is performed. Moreover, the derived algorithm, unlike the conventional better estimates free of the initialEM algorithm, can obtain parameter values.