A broadly applicable algorithm for computing maximum likelihood estimates from incomplete data is presented at various levels of generality. Theory showing the monotone behaviour of the likelihood and ...

This is a preview. Log in through your library . Abstract Two major reasons for the popularity of the EM algorithm are that its maximum step involves only complete-data maximum likelihood estimation, ...

The previous implementation of the simulated annealing algorithm in Colony program did not allow for extensive searches, to save computational time, for the maximum likelihood configurations.