Package jsat.math.optimization.stochastic

Interface Summary

Interface	Description
GradientUpdater	This interface defines the method of updating some weight vector using a gradient and a learning rate.

Class Summary

Class	Description
AdaDelta	AdaDelta is inspired by AdaGrad and was developed for use primarily in neural networks.
AdaGrad	AdaGrad provides an adaptive learning rate for each individual feature See: Duchi, J., Hazan, E.,&Singer, Y.
Adam	Adam is inspired by RMSProp and AdaGrad, where the former can be seen as a special case of Adam.
NAdaGrad	Normalized AdaGrad provides an adaptive learning rate for each individual feature, and is mostly scale invariant to the data distribution.
RMSProp	rmsprop is an adpative learning weight scheme proposed by Geoffrey Hinton.
Rprop	The Rprop algorithm provides adaptive learning rates using only first order information.
SGDMomentum	Performs unaltered Stochastic Gradient Decent updates using either standard or Nestrov momentum.
SimpleSGD	Performs unaltered Stochastic Gradient Decent updates computing x = x- η grad Because the SimpleSGD requires no internal state, it is not necessary to call SimpleSGD.setup(int).