Wolfram Language & System 11.0 (2016)|Legacy Documentation

This is documentation for an earlier version of the Wolfram Language.View current documentation (Version 11.2)

CrossEntropyLossLayer

CrossEntropyLossLayer[]
represents a net layer that computes the cross-entropy loss by comparing probabilities with specified target values.

CrossEntropyLossLayer[form]
specifies the form of target that has been supplied.

Details and OptionsDetails and Options

  • In CrossEntropyLossLayer[form], possible forms include:
  • "Index"single integer giving the index of the target result
    "Probabilities"vector of target probabilities for different possible results
  • CrossEntropyLossLayer[][<|"Input"->in,"Target"target|>] explicitly computes the output from applying the layer.
  • CrossEntropyLossLayer[][<|"Input"->{in1,in2,},"Target"->{target1,target2,}|>] explicitly computes outputs for each of the ini and targeti.
  • CrossEntropyLossLayer exposes the following ports for use in NetGraph etc.:
  • "Input"a numeric vector
    "Target"an integer if sparse is True and a numeric vector if False
    "Loss"a real number
  • CrossEntropyLossLayer is typically used inside NetChain, NetGraph, and NetTrain.
  • When it cannot be inferred from other layers in a larger net, the option "Input"->n can be used to fix the input dimensions of CrossEntropyLossLayer.

ExamplesExamplesopen allclose all

Basic Examples  (4)Basic Examples  (4)

Create a CrossEntropyLossLayer object:

In[1]:=
Click for copyable input
Out[1]=

Apply data to a CrossEntropyLossLayer, where the label is the index of the target class:

In[1]:=
Click for copyable input
In[2]:=
Click for copyable input
Out[2]=

Apply data to a CrossEntropyLossLayer, where the label is a vector of class probabilities:

In[1]:=
Click for copyable input
In[2]:=
Click for copyable input
Out[2]=

Create a CrossEntropyLossLayer that takes in True or False as the target using a NetEncoder:

In[1]:=
Click for copyable input
Out[1]=

Apply the layer to an example:

In[2]:=
Click for copyable input
Out[2]=
Introduced in 2016
(11.0)