Poisson

public class Poisson

Computes the Poisson loss between labels and predictions.

loss = predictions - labels * log(predictions)

Standalone usage:

    Operand<TFloat32> labels =
        tf.constant(new float[][] { {0.f, 1.f}, {0.f, 0.f} });
    Operand<TFloat32> predictions =
        tf.constant(new float[][] { {1.f, 1.f}, {0.f, 0.f} });
    Poisson poissonLoss = new Poisson(tf);
    Operand<TFloat32> result = poissonLoss.call(labels, predictions);
    // produces 0.5f
 

Calling with sample weight:

    Operand<TFloat32> sampleWeight = tf.constant(new float[] {0.8f, 0.2f});
    Operand<TFloat32> result = poissonLoss.call(labels, predictions, sampleWeight);
    // produces 0.4f
 

Using SUM reduction type:

    Poisson poissonLoss = new Poisson(tf, Reduction.SUM);
    Operand<TFloat32> result = poissonLoss.call(labels, predictions);
    // produces 0.999f
 

Using NONE reduction type:

    Poisson poissonLoss = new Poisson(tf, Reduction.NONE);
    Operand<TFloat32> result = poissonLoss.call(labels, predictions);
    // produces [0.999f, 0f]
 

Inherited Fields

Public Constructors

Poisson(Ops tf)
Creates a Poisson Loss using getSimpleName() as the loss name and a Loss Reduction of REDUCTION_DEFAULT
Poisson(Ops tf, String name)
Creates a Poisson Loss using a Loss Reduction of REDUCTION_DEFAULT
Poisson(Ops tf, Reduction reduction)
Creates a Poisson Loss using getSimpleName() as the loss name
Poisson(Ops tf, String name, Reduction reduction)
Creates a Poisson Loss

Public Methods

<T extends TNumber> Operand<T>
call(Operand<? extends TNumber> labels, Operand<T> predictions, Operand<T> sampleWeights)
Generates an Operand that calculates the loss.

Inherited Methods

Public Constructors

public Poisson (Ops tf)

Creates a Poisson Loss using getSimpleName() as the loss name and a Loss Reduction of REDUCTION_DEFAULT

Parameters
tf the TensorFlow Ops

public Poisson (Ops tf, String name)

Creates a Poisson Loss using a Loss Reduction of REDUCTION_DEFAULT

Parameters
tf the TensorFlow Ops
name the name of the loss, if null then getSimpleName() is used.

public Poisson (Ops tf, Reduction reduction)

Creates a Poisson Loss using getSimpleName() as the loss name

Parameters
tf the TensorFlow Ops
reduction Type of Reduction to apply to the loss.

public Poisson (Ops tf, String name, Reduction reduction)

Creates a Poisson Loss

Parameters
tf the TensorFlow Ops
name the name of the loss, if null then getSimpleName() is used.
reduction Type of Reduction to apply to the loss.

Public Methods

public Operand<T> call (Operand<? extends TNumber> labels, Operand<T> predictions, Operand<T> sampleWeights)

Generates an Operand that calculates the loss.

Parameters
labels the truth values or labels
predictions the predictions
sampleWeights Optional sampleWeights acts as a coefficient for the loss. If a scalar is provided, then the loss is simply scaled by the given value. If SampleWeights is a tensor of size [batch_size], then the total loss for each sample of the batch is rescaled by the corresponding element in the SampleWeights vector. If the shape of SampleWeights is [batch_size, d0, .. dN-1] (or can be broadcast to this shape), then each loss element of predictions is scaled by the corresponding value of SampleWeights. (Note on dN-1: all loss functions reduce by 1 dimension, usually axis=-1.)
Returns
  • the loss