Module nipals

source code

The :mod:`parsimony.algorithms.nipals` module includes several algorithms that minimises an implicit loss function based on the NIPALS algorithm.

Algorithms may not store states. I.e., if they are classes, do not keep references to objects with state in the algorithm objects. It should be possible to copy and share algorithms between e.g. estimators, and thus they should not depend on any state.

Created on Thu Feb 20 17:46:17 2014

Copyright (c) 2013-2014, CEA/DSV/I2BM/Neurospin. All rights reserved.