A Metric that computes the categorical cross-entropy loss between true labels and predicted labels.
This is the crossentropy metric class to be used when there are multiple label classes (2 or
more). The labels should be given as a one_hot representation. eg., When labels values are
[2, 0, 1]
, the labels Operand contains = [[0, 0, 1], [1, 0, 0], [0, 1, 0]]
.
Inherited Constants
Public Constructors
CategoricalCrossentropy(Ops tf, String name, boolean fromLogits, float labelSmoothing, long seed, Class<T> type)
Creates a CategoricalCrossentropy metric that computes the crossentropy metric between the
labels and predictions.
|
|
CategoricalCrossentropy(Ops tf, String name, boolean fromLogits, float labelSmoothing, int axis, long seed, Class<T> type)
Creates a CategoricalCrossentropy metric that computes the crossentropy metric between the
labels and predictions.
|
Public Methods
Inherited Methods
LossMetric<T> |
getLoss()
Gets the loss function.
|
List<Op> |
updateStateList(Operand<? extends TNumber> labels, Operand<? extends TNumber> predictions, Operand<? extends TNumber> sampleWeights)
Creates Operations that update the state of the mean metric, by calling the loss function and
passing the loss to the Mean metric to calculate the weighted mean of the loss over many
iterations.
|
Variable<T> |
getCount()
Gets the count variable
|
Class<T> |
getResultType()
Gets the type for the variables
|
Variable<T> |
getTotal()
Gets the total variable
|
Op |
resetStates()
Resets any state variables to their initial values
|
Operand<T> |
result()
Gets the current result of the metric
|
List<Op> |
updateStateList(Operand<? extends TNumber> values, Operand<? extends TNumber> sampleWeights)
Updates the metric variables based on the inputs.
|
final Operand<T> | |
String |
getName()
Gets the name of this metric.
|
long |
getSeed()
Gets the random number generator seed value
|
Ops |
getTF()
Gets the TensorFlow Ops
|
abstract Op |
resetStates()
Resets any state variables to their initial values
|
abstract Operand<T> |
result()
Gets the current result of the metric
|
final Op | |
final Op |
updateState(Operand<? extends TNumber> values, Operand<? extends TNumber> sampleWeights)
Creates a NoOp Operation with control dependencies to update the metric state
|
List<Op> | |
List<Op> |
updateStateList(Operand<? extends TNumber> values, Operand<? extends TNumber> sampleWeights)
Creates a List of Operations to update the metric state based on input values.
|
boolean |
equals(Object arg0)
|
final Class<?> |
getClass()
|
int |
hashCode()
|
final void |
notify()
|
final void |
notifyAll()
|
String |
toString()
|
final void |
wait(long arg0, int arg1)
|
final void |
wait(long arg0)
|
final void |
wait()
|
Public Constructors
public CategoricalCrossentropy (Ops tf, String name, boolean fromLogits, float labelSmoothing, long seed, Class<T> type)
Creates a CategoricalCrossentropy metric that computes the crossentropy metric between the labels and predictions.
Uses a CHANNELS_LAST
for the channel axis.
Parameters
tf | the TensorFlow Ops |
---|---|
name | the name of this metric, if null then metric name is getSimpleName() . |
fromLogits | Whether to interpret predictions as a tensor of logit values oras opposed to a probability distribution. |
labelSmoothing | value used to smooth labels, When > 0, label values are smoothed,
meaning the confidence on label values are relaxed. e.g. labelSmoothing=0.2
means that we will use a value of 0.1 for label 0 and 0.9
for label 1 |
seed | the seed for random number generation. An initializer created with a given seed will always produce the same random tensor for a given shape and data type. |
type | the type for the variables and result |
public CategoricalCrossentropy (Ops tf, String name, boolean fromLogits, float labelSmoothing, int axis, long seed, Class<T> type)
Creates a CategoricalCrossentropy metric that computes the crossentropy metric between the labels and predictions.
Parameters
tf | the TensorFlow Ops |
---|---|
name | the name of this metric, if null then metric name is getSimpleName() . |
fromLogits | Whether to interpret predictions as a tensor of logit values as opposed to a probability distribution. |
labelSmoothing | value used to smooth labels, When > 0, label values are smoothed,
meaning the confidence on label values are relaxed. e.g. labelSmoothing=0.2
means that we will use a value of 0.1 for label 0 and 0.9
for label 1 |
axis | Int specifying the channels axis. axis=
corresponds to data format channels_last , and
axis= corresponds to data format
channels_first . |
seed | the seed for random number generation. An initializer created with a given seed will always produce the same random tensor for a given shape and data type. |
type | the type for the variables and result |