>  Docs Center  >  IDL Reference  >  Machine Learning  >  IDLmllfHuber

IDLmllfHuber

IDLmllfHuber

In machine learning, a loss function is a mathematical function that must be minimized in order to achieve convergence. Choosing the proper loss function is an important step in designing your neural network. The IDLmllfHuber (Huber) loss function is implemented with the following formula:

where x is the calculated output of the model and y is the predicted output or truth.

Example


Compile_opt idl2
LossFunction = IDLmllfHuber(0.5)
Print, LossFunction(Findgen(10)/9.0, Fltarr(10))
 

Typically, you will pass an object of this class to a neural network model definition:

Classifier = IDLmlFeedForwardNeuralNetwork([3, 7, 1],
LOSS_FUNCTION=IDLmllfHuber(0.25)

Syntax


Kernel = IDLmllfHuber(Delta)

Arguments


Delta

Specify a threshold value that determines how small the error has to be to make it quadratic.

Keywords


None

Version History


8.7.1

Introduced

See Also


IDLmllfCrossEntropy, IDLmllfLogCosh, IDLmllfMeanAbsoluteError, IDLmllfMeanSquaredError



© 2019 Harris Geospatial Solutions, Inc. |  Legal
My Account    |    Store    |    Contact Us