Termination criteria for EM algorithms.
Expectation maximization iteratively produces an estimate of sufficient statistics for learnable parameters, then maximizes the parameters according to the estimate.
Expectation maximization iteratively produces an estimate of sufficient statistics for learnable parameters, then maximizes the parameters according to the estimate. This trait can be extended with a different expectation or maximization algorithm; see the code for details.
An EM algorithm which learns parameters using a factored algorithm
An EM algorithm which learns parameters using an inference algorithm provided as an argument
An EM algorithm which learns parameters using an inference algorithm provided as an argument
Terminate when the maximum number of iterations has been reached
An EM algorithm which learns parameters incrementally
An online EM algorithm which learns parameters using a factored algorithm
Methods for creating probabilistic factors associated with elements and their sufficient statistics.
Terminate when the magnitude of sufficient statistics does not exhibit a change greater than the specified tolerance.
Termination criteria for EM algorithms. A termination criteria can be passed as an argument to the EM apply method.