This class implements AMF (alternating matrix factorization) on the given matrix V.
Alternating matrix factorization decomposes V in the form where W is called the basis matrix and H is called the encoding matrix. V is taken to be of size n x m and the obtained W is n x r and H is r x m. The size r is called the rank of the factorization.
The implementation requires three template types; the first contains the policy used to determine when the algorithm has converged; the second contains the initialization rule for the W and H matrix; the last contains the update rule to be used during each iteration. This templatization allows the user to try various update rules, initialization rules, and termination policies (including ones not supplied with mlpack) for factorization. By default, the template parameters to AMF implement non-negative matrix factorization with the multiplicative distance update.
A simple example of how to run AMF (or NMF) is shown below.
extern arma::mat V; // Matrix that we want to perform LMF on.
Create the AMF object and (optionally) set the parameters which AMF will run with.
The minimum residue refers to the root mean square of the difference between two subsequent iterations of the product W * H. A low residue indicates that subsequent iterations are not producing much change in W and H. Once the residue goes below the specified minimum residue, the algorithm terminates.
Parameters
initializeRule
Optional instantiated InitializationRule object for initializing the W and H matrices.
update
Optional instantiated UpdateRule object; this parameter is useful when the update rule for the W and H vector has state that it needs to store (i.e. HUpdate() and WUpdate() are not static functions).
