NN - org.platanios.tensorflow.api.ops.NN

Type Members

sealed trait CNNDataFormat extends AnyRef

Definition Classes
NN
sealed trait ConvPaddingMode extends AnyRef

Padding mode.
case class NNOps(output: Output) extends Product with Serializable

Value Members

final def !=(arg0: Any): Boolean

Definition Classes
AnyRef → Any
final def ##(): Int

Definition Classes
AnyRef → Any
final def ==(arg0: Any): Boolean

Definition Classes
AnyRef → Any
object CNNDataFormat

Definition Classes
NN
object ConvPaddingMode
object NCWFormat extends CNNDataFormat with Product with Serializable

Definition Classes
NN
object NWCFormat extends CNNDataFormat with Product with Serializable

Definition Classes
NN
object SameConvPadding extends ConvPaddingMode with Product with Serializable

Definition Classes
NN
object ValidConvPadding extends ConvPaddingMode with Product with Serializable

Definition Classes
NN
def addBias(value: Output, bias: Output, cNNDataFormat: CNNDataFormat = CNNDataFormat.default, name: String = "AddBias"): Output

$OpDocNNAddBias
$OpDocNNAddBias
value
Value tensor.
bias
Bias tensor that must be one-dimensional (i.e., it must have rank 1).
cNNDataFormat
Data format of the input and output tensors. With the default format NWCFormat, the bias tensor will be added to the last dimension of the value tensor. Alternatively, the format could be NCWFormat, and the bias tensor would be added to the third-to-last dimension.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
final def asInstanceOf[T0]: T0

Definition Classes
Any
def batchNormalization(x: Output, mean: Output, variance: Output, offset: Option[Output] = None, scale: Option[Output] = None, epsilon: Output, name: String = "BatchNormalization"): Output

$OpDocNNBatchNormalization
$OpDocNNBatchNormalization
x
Input tensor of arbitrary dimensionality.
mean
Mean tensor.
variance
Variance tensor.
offset
Optional offset tensor, often denoted beta in equations.
scale
Optional scale tensor, often denoted gamma in equations.
epsilon
Small floating point number added to the variance to avoid division by zero.
name
Name for the created ops.
returns
Batch-normalized tensor x.

Definition Classes
NN
def clone(): AnyRef

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( ... )
def conv2D(input: Output, filter: Output, stride1: Long, stride2: Long, padding: ConvPaddingMode, dataFormat: CNNDataFormat = CNNDataFormat.default, dilations: (Int, Int, Int, Int) = (1, 1, 1, 1), useCuDNNOnGPU: Boolean = true, name: String = "Conv2D"): Output

$OpDocNNConv2D
$OpDocNNConv2D
input
4-D tensor whose dimension order is interpreted according to the value of dataFormat.
filter
4-D tensor with shape [filterHeight, filterWidth, inChannels, outChannels].
stride1
Stride of the sliding window along the second dimension of input.
stride2
Stride of the sliding window along the third dimension of input.
padding
Padding mode to use.
dataFormat
Format of the input and output data.
dilations
The dilation factor for each dimension of input. If set to k > 1, there will be k - 1 skipped cells between each filter element on that dimension. The dimension order is determined by the value of dataFormat. Dilations in the batch and depth dimensions must be set to 1.
useCuDNNOnGPU
Boolean value indicating whether or not to use CuDNN for the created op, if its placed on a GPU, as opposed to the TensorFlow implementation.
name
Name for the created op.
returns
Created op output, which is a 4-D tensor whose dimension order depends on the value of dataFormat.

Definition Classes
NN
def conv2DBackpropFilter(input: Output, filterSizes: Output, outputGradient: Output, stride1: Long, stride2: Long, padding: ConvPaddingMode, dataFormat: CNNDataFormat = CNNDataFormat.default, dilations: (Int, Int, Int, Int) = (1, 1, 1, 1), useCuDNNOnGPU: Boolean = true, name: String = "Conv2DBackpropFilter"): Output

$OpDocNNConv2DBackpropFilter
$OpDocNNConv2DBackpropFilter
input
4-D tensor whose dimension order is interpreted according to the value of dataFormat.
filterSizes
Integer vector representing the shape of the original filter, which is a 4-D tensor.
outputGradient
4-D tensor containing the gradients w.r.t. the output of the convolution and whose shape depends on the value of dataFormat.
stride1
Stride of the sliding window along the second dimension of input.
stride2
Stride of the sliding window along the third dimension of input.
padding
Padding mode to use.
dataFormat
Format of the input and output data.
dilations
The dilation factor for each dimension of input. If set to k > 1, there will be k - 1 skipped cells between each filter element on that dimension. The dimension order is determined by the value of dataFormat. Dilations in the batch and depth dimensions must be set to 1.
useCuDNNOnGPU
Boolean value indicating whether or not to use CuDNN for the created op, if its placed on a GPU, as opposed to the TensorFlow implementation.
name
Name for the created op.
returns
Created op output, which is a 4-D tensor whose dimension order depends on the value of dataFormat.

Definition Classes
NN
def conv2DBackpropInput(inputSizes: Output, filter: Output, outputGradient: Output, stride1: Long, stride2: Long, padding: ConvPaddingMode, dataFormat: CNNDataFormat = CNNDataFormat.default, dilations: (Int, Int, Int, Int) = (1, 1, 1, 1), useCuDNNOnGPU: Boolean = true, name: String = "Conv2DBackpropInput"): Output

$OpDocNNConv2DBackpropInput
$OpDocNNConv2DBackpropInput
inputSizes
Integer vector representing the shape of the original input, which is a 4-D tensor.
filter
4-D tensor with shape [filterHeight, filterWidth, inChannels, outChannels].
outputGradient
4-D tensor containing the gradients w.r.t. the output of the convolution and whose shape depends on the value of dataFormat.
stride1
Stride of the sliding window along the second dimension of input.
stride2
Stride of the sliding window along the third dimension of input.
padding
Padding mode to use.
dataFormat
Format of the input and output data.
dilations
The dilation factor for each dimension of input. If set to k > 1, there will be k - 1 skipped cells between each filter element on that dimension. The dimension order is determined by the value of dataFormat. Dilations in the batch and depth dimensions must be set to 1.
useCuDNNOnGPU
Boolean value indicating whether or not to use CuDNN for the created op, if its placed on a GPU, as opposed to the TensorFlow implementation.
name
Name for the created op.
returns
Created op output, which is a 4-D tensor whose dimension order depends on the value of dataFormat.

Definition Classes
NN
def crelu(input: Output, axis: Output = 1, name: String = "CReLU"): Output

$OpDocNNCrelu
$OpDocNNCrelu
input
Input tensor.
axis
Along along which the output values are concatenated along.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def dropout(input: Output, keepProbability: Float, scaleOutput: Boolean = true, noiseShape: Output = null, seed: Option[Int] = None, name: String = "Dropout"): Output

$OpDocNNDropout
$OpDocNNDropout
input
Input tensor.
keepProbability
Probability (i.e., number in the interval (0, 1]) that each element is kept.
scaleOutput
If true, the outputs will be divided by the keep probability.
noiseShape
INT32 rank-1 tensor representing the shape for the randomly generated keep/drop flags.
seed
Optional random seed, used to generate a random seed pair for the random number generator, when combined with the graph-level seed.
name
Name for the created op.
returns
Created op output that has the same shape as input.

Definition Classes
NN
def dynamicDropout(input: Output, keepProbability: Output, scaleOutput: Boolean = true, noiseShape: Output = null, seed: Option[Int] = None, name: String = "Dropout"): Output

$OpDocNNDropout
$OpDocNNDropout
input
Input tensor.
keepProbability
Probability (i.e., scalar in the interval (0, 1]) that each element is kept.
scaleOutput
If true, the outputs will be divided by the keep probability.
noiseShape
INT32 rank-1 tensor representing the shape for the randomly generated keep/drop flags.
seed
Optional random seed, used to generate a random seed pair for the random number generator, when combined with the graph-level seed.
name
Name for the created op.
returns
Created op output that has the same shape as input.

Definition Classes
NN
def elu[T](x: T, name: String = "ELU")(implicit arg0: OutputOps[T]): T

$OpDocNNElu
$OpDocNNElu
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
final def eq(arg0: AnyRef): Boolean

Definition Classes
AnyRef
def equals(arg0: Any): Boolean

Definition Classes
AnyRef → Any
def finalize(): Unit

Attributes
protected[java.lang]
Definition Classes
AnyRef
Annotations
@throws( classOf[java.lang.Throwable] )
def fusedBatchNormalization(x: Output, scale: Output, offset: Output, mean: Option[Output] = None, variance: Option[Output] = None, epsilon: Float = 0.001f, dataFormat: CNNDataFormat = NWCFormat, isTraining: Boolean = true, name: String = "FusedBatchNormalization"): (Output, Output, Output)

$OpDocNNFusedBatchNormalization
$OpDocNNFusedBatchNormalization
x
Input tensor with 4 dimensions.
scale
Vector used for scaling.
offset
Vector used as an added offset.
mean
Optional population mean vector, used for inference only.
variance
Optional population variance vector, used for inference only.
epsilon
Small floating point number added to the variance to avoid division by zero.
dataFormat
Data format for x.
isTraining
Boolean value indicating whether the operation is used for training or inference.
name
Name for the created ops.
returns
Batch normalized tensor x, along with the a batch mean vector, and a batch variance vector.

Definition Classes
NN
final def getClass(): Class[_]

Definition Classes
AnyRef → Any
def hashCode(): Int

Definition Classes
AnyRef → Any
def inTopK(predictions: Output, targets: Output, k: Output, name: String = "InTopK"): Output

$OpDocNNInTopK
$OpDocNNInTopK
predictions
FLOAT32 tensor containing the predictions.
targets
INT32 or INT64 tensor containing the targets.
k
Scalar INT32 or INT64 tensor containing the number of top elements to look at.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
final def isInstanceOf[T0]: Boolean

Definition Classes
Any
def l2Loss(input: Output, name: String = "L2Loss"): Output

$OpDocNNL2Loss
$OpDocNNL2Loss
input
FLOAT16, FLOAT32, or FLOAT64 input tensor.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def l2Normalize(x: Output, axes: Output, epsilon: Float = 1e-12f, name: String = "L2Normalize"): Output

$OpDocNNL2Normalize
$OpDocNNL2Normalize
x
Input tensor.
axes
Tensor containing the axes along which to normalize.
epsilon
Lower bound value for the norm. The created op will use sqrt(epsilon) as the divisor, if norm < sqrt(epsilon).
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def linear(x: Output, weights: Output, bias: Output = null, name: String = "Linear"): Output

$OpDocNNLinear
$OpDocNNLinear
x
Input tensor.
weights
Weights tensor.
bias
Bias tensor.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def localResponseNormalization(input: Output, depthRadius: Int = 5, bias: Float = 1.0f, alpha: Float = 1.0f, beta: Float = 0.5f, name: String = "LocalResponseNormalization"): Output

$OpDocNNLocalResponseNormalization
$OpDocNNLocalResponseNormalization
input
Input tensor with data type FLOAT16, BFLOAT16, or FLOAT32.
depthRadius
Half-width of the 1-D normalization window.
bias
Offset (usually positive to avoid dividing by 0).
alpha
Scale factor (usually positive).
beta
Exponent.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def logPoissonLoss(logPredictions: Output, targets: Output, computeFullLoss: Boolean = false, name: String = "LogPoissonLoss"): Output

$OpDocNNLogPoissonLoss
$OpDocNNLogPoissonLoss
logPredictions
Tensor containing the log-predictions.
targets
Tensor with the same shape as logPredictions, containing the target values.
computeFullLoss
If true, Stirling's Approximation is used to approximate the full loss. Defaults to false, meaning that the constant term is ignored.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def logSoftmax(logits: Output, axis: Int = 1, name: String = "LogSoftmax"): Output

$OpDocNNLogSoftmax
$OpDocNNLogSoftmax
logits
Tensor containing the logits with data type FLOAT16, FLOAT32, or FLOAT64.
axis
Axis along which to perform the log-softmax. Defaults to -1 denoting the last axis.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def lrn(input: Output, depthRadius: Int = 5, bias: Float = 1.0f, alpha: Float = 1.0f, beta: Float = 0.5f, name: String = "LRN"): Output

$OpDocNNLocalResponseNormalization
$OpDocNNLocalResponseNormalization
input
Input tensor with data type FLOAT16, BFLOAT16, or FLOAT32.
depthRadius
Half-width of the 1-D normalization window.
bias
Offset (usually positive to avoid dividing by 0).
alpha
Scale factor (usually positive).
beta
Exponent.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def maxPool(input: Output, windowSize: Seq[Long], stride1: Long, stride2: Long, padding: ConvPaddingMode, dataFormat: CNNDataFormat = CNNDataFormat.default, name: String = "MaxPool"): Output

$OpDocNNMaxPool
$OpDocNNMaxPool
input
4-D tensor whose dimension order is interpreted according to the value of dataFormat.
windowSize
The size of the pooling window for each dimension of the input tensor.
stride1
Stride of the sliding window along the second dimension of input.
stride2
Stride of the sliding window along the third dimension of input.
padding
Padding mode to use.
dataFormat
Format of the input and output data.
name
Name for the created op.
returns
Created op output, which is a 4-D tensor whose dimension order depends on the value of dataFormat.

Definition Classes
NN
def maxPoolGrad(originalInput: Output, originalOutput: Output, outputGradient: Output, windowSize: Seq[Long], stride1: Long, stride2: Long, padding: ConvPaddingMode, dataFormat: CNNDataFormat = CNNDataFormat.default, name: String = "MaxPoolGrad"): Output

$OpDocNNMaxPoolGrad
$OpDocNNMaxPoolGrad
originalInput
Original input tensor.
originalOutput
Original output tensor.
outputGradient
4-D tensor containing the gradients w.r.t. the output of the max pooling and whose shape depends on the value of dataFormat.
windowSize
The size of the pooling window for each dimension of the input tensor.
stride1
Stride of the sliding window along the second dimension of input.
stride2
Stride of the sliding window along the third dimension of input.
padding
Padding mode to use.
dataFormat
Format of the input and output data.
name
Name for the created op.
returns
Created op output, which is a 4-D tensor whose dimension order depends on the value of dataFormat.

Definition Classes
NN
def maxPoolGradGrad(originalInput: Output, originalOutput: Output, outputGradient: Output, windowSize: Seq[Long], stride1: Long, stride2: Long, padding: ConvPaddingMode, dataFormat: CNNDataFormat = CNNDataFormat.default, name: String = "MaxPoolGradGrad"): Output

$OpDocNNMaxPoolGradGrad
$OpDocNNMaxPoolGradGrad
originalInput
Original input tensor.
originalOutput
Original output tensor.
outputGradient
4-D tensor containing the gradients w.r.t. the output of the max pooling and whose shape depends on the value of dataFormat.
windowSize
The size of the pooling window for each dimension of the input tensor.
stride1
Stride of the sliding window along the second dimension of input.
stride2
Stride of the sliding window along the third dimension of input.
padding
Padding mode to use.
dataFormat
Format of the input and output data.
name
Name for the created op.
returns
Created op output, which is a 4-D tensor whose dimension order depends on the value of dataFormat.

Definition Classes
NN
final def ne(arg0: AnyRef): Boolean

Definition Classes
AnyRef
final def notify(): Unit

Definition Classes
AnyRef
final def notifyAll(): Unit

Definition Classes
AnyRef
def relu(input: Output, alpha: Float = 0.0f, name: String = "ReLU"): Output

$OpDocNNRelu
$OpDocNNRelu
input
Input tensor.
alpha
Slope of the negative section, also known as leakage parameter. If other than 0.0f, the negative part will be equal to alpha * x instead of 0. Defaults to 0.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def relu6[T](x: T, name: String = "ReLU6")(implicit arg0: OutputOps[T]): T

$OpDocNNRelu6
$OpDocNNRelu6
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def selu[T](x: T, name: String = "SELU")(implicit arg0: OutputOps[T]): T

$OpDocNNSelu
$OpDocNNSelu
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def sequenceLoss(logits: Output, labels: Output, weights: Output = null, averageAcrossTimeSteps: Boolean = true, averageAcrossBatch: Boolean = true, lossFn: (Output, Output) ⇒ Output = sparseSoftmaxCrossEntropy(_, _), name: String = "SequenceLoss"): Output

$OpDocNNSequenceLoss
$OpDocNNSequenceLoss
logits
Tensor of shape [batchSize, sequenceLength, numClasses] containing unscaled log probabilities.
labels
Tensor of shape [batchSize, sequenceLength] containing the true label at each time step.
weights
Optionally, a tensor of shape [batchSize, sequenceLength] containing weights to use for each prediction. When using weights as masking, set all valid time steps to 1 and all padded time steps to 0 (e.g., a mask returned by tf.sequenceMask).
averageAcrossTimeSteps
If true, the loss is summed across the sequence dimension and divided by the total label weight across all time steps.
averageAcrossBatch
If true, the loss is summed across the batch dimension and divided by the batch size.
lossFn
Loss function to use that takes the predicted logits and the true labels as inputs and returns the loss value. Defaults to sparseSoftmaxCrossEntropy.
name
Name prefix to use for the created ops.
returns
Created op output.

Definition Classes
NN
Annotations
@throws( ... )
Exceptions thrown
InvalidShapeException If any of logits, labels, or weights has invalid shape.
def sigmoidCrossEntropy(logits: Output, labels: Output, weights: Output = null, name: String = "SigmoidCrossEntropy"): Output

$OpDocNNSigmoidCrossEntropy
$OpDocNNSigmoidCrossEntropy
logits
Tensor of shape [D0, D1, ..., Dr-1, numClasses] and data type FLOAT16, FLOAT32, or FLOAT64, containing unscaled log probabilities.
labels
Tensor of shape [D0, D1, ..., Dr-1, numClasses] and data type FLOAT16, FLOAT32, or FLOAT64, where each row must be a valid probability distribution.
weights
Optionally, a coefficient to use for the positive examples.
name
Name for the created op.
returns
Created op output, with rank one less than that of logits and the same data type as logits, containing the sigmoid cross entropy loss.

Definition Classes
NN
def softmax(logits: Output, axis: Int = 1, name: String = "Softmax"): Output

$OpDocNNSoftmax
$OpDocNNSoftmax
logits
Tensor containing the logits with data type FLOAT16, FLOAT32, or FLOAT64.
axis
Axis along which to perform the softmax. Defaults to -1 denoting the last axis.
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def softmaxCrossEntropy(logits: Output, labels: Output, axis: Int = 1, name: String = "SoftmaxCrossEntropy"): Output

$OpDocNNSoftmaxCrossEntropy
$OpDocNNSoftmaxCrossEntropy
logits
Tensor of shape [D0, D1, ..., Dr-1, numClasses] and data type FLOAT16, FLOAT32, or FLOAT64, containing unscaled log probabilities.
labels
Tensor of shape [D0, D1, ..., Dr-1, numClasses] and data type FLOAT16, FLOAT32, or FLOAT64, where each row must be a valid probability distribution.
axis
The class axis, along which the softmax is computed. Defaults to -1, which is the last axis.
name
Name for the created op.
returns
Created op output, with rank one less than that of logits and the same data type as logits, containing the softmax cross entropy loss.

Definition Classes
NN
def softplus[T](x: T, name: String = "Softplus")(implicit arg0: OutputOps[T]): T

$OpDocNNSoftplus
$OpDocNNSoftplus
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def softsign[T](x: T, name: String = "Softsign")(implicit arg0: OutputOps[T]): T

$OpDocNNSoftsign
$OpDocNNSoftsign
name
Name for the created op.
returns
Created op output.

Definition Classes
NN
def sparseSoftmaxCrossEntropy(logits: Output, labels: Output, axis: Int = 1, name: String = "SparseSoftmaxCrossEntropy"): Output

$OpDocNNSparseSoftmaxCrossEntropy
$OpDocNNSparseSoftmaxCrossEntropy
logits
Tensor of shape [D0, D1, ..., Dr-1, numClasses] (where r is the rank of labels and of the result) and data type FLOAT16, FLOAT32, or FLOAT64, containing unscaled log probabilities.
labels
Tensor of shape [D0, D1, ..., Dr-1] (where r is the rank of labels and of the result) and data type INT32 or INT64. Each entry in labels must be an index in [0, numClasses). Other values will raise an exception when this op is run on a CPU, and return NaN values for the corresponding loss and gradient rows when this op is run on a GPU.
axis
The class axis, along which the softmax is computed. Defaults to -1, which is the last axis.
name
Name for the created op.
returns
Created op output, with the same shape as labels and the same data type as logits, containing the softmax cross entropy loss.

Definition Classes
NN
final def synchronized[T0](arg0: ⇒ T0): T0

Definition Classes
AnyRef
def toString(): String

Definition Classes
AnyRef → Any
def topK(input: Output, k: Output = 1, sorted: Boolean = true, name: String = "TopK"): (Output, Output)

$OpDocNNTopK
$OpDocNNTopK
input
Input tensor whose last axis has size at least k.
k
Scalar INT32 tensor containing the number of top elements to look for along the last axis of input.
sorted
If true, the resulting k elements will be sorted by their values in descending order.
name
Name for the created op.
returns
Tuple containing the created op outputs: (i) values: the k largest elements along each last dimensional slice, and (ii) indices: the indices of values within the last axis of input.

Definition Classes
NN
final def wait(): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
final def wait(arg0: Long, arg1: Int): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )
final def wait(arg0: Long): Unit

Definition Classes
AnyRef
Annotations
@throws( ... )

NN

Related Doc: package ops

object NN extends NN

Type Members

sealed trait CNNDataFormat extends AnyRef

sealed trait ConvPaddingMode extends AnyRef

case class NNOps(output: Output) extends Product with Serializable

Value Members

final def !=(arg0: Any): Boolean

final def ##(): Int

final def ==(arg0: Any): Boolean

object CNNDataFormat

object ConvPaddingMode

object NCWFormat extends CNNDataFormat with Product with Serializable

object NWCFormat extends CNNDataFormat with Product with Serializable

object SameConvPadding extends ConvPaddingMode with Product with Serializable

object ValidConvPadding extends ConvPaddingMode with Product with Serializable

def addBias(value: Output, bias: Output, cNNDataFormat: CNNDataFormat = CNNDataFormat.default, name: String = "AddBias"): Output

final def asInstanceOf[T0]: T0

def batchNormalization(x: Output, mean: Output, variance: Output, offset: Option[Output] = None, scale: Option[Output] = None, epsilon: Output, name: String = "BatchNormalization"): Output

def clone(): AnyRef

def conv2D(input: Output, filter: Output, stride1: Long, stride2: Long, padding: ConvPaddingMode, dataFormat: CNNDataFormat = CNNDataFormat.default, dilations: (Int, Int, Int, Int) = (1, 1, 1, 1), useCuDNNOnGPU: Boolean = true, name: String = "Conv2D"): Output

def crelu(input: Output, axis: Output = 1, name: String = "CReLU"): Output

def dropout(input: Output, keepProbability: Float, scaleOutput: Boolean = true, noiseShape: Output = null, seed: Option[Int] = None, name: String = "Dropout"): Output

def dynamicDropout(input: Output, keepProbability: Output, scaleOutput: Boolean = true, noiseShape: Output = null, seed: Option[Int] = None, name: String = "Dropout"): Output

def elu[T](x: T, name: String = "ELU")(implicit arg0: OutputOps[T]): T

final def eq(arg0: AnyRef): Boolean

def equals(arg0: Any): Boolean

def finalize(): Unit

def fusedBatchNormalization(x: Output, scale: Output, offset: Output, mean: Option[Output] = None, variance: Option[Output] = None, epsilon: Float = 0.001f, dataFormat: CNNDataFormat = NWCFormat, isTraining: Boolean = true, name: String = "FusedBatchNormalization"): (Output, Output, Output)

final def getClass(): Class[_]

def hashCode(): Int

def inTopK(predictions: Output, targets: Output, k: Output, name: String = "InTopK"): Output

final def isInstanceOf[T0]: Boolean

def l2Loss(input: Output, name: String = "L2Loss"): Output

def l2Normalize(x: Output, axes: Output, epsilon: Float = 1e-12f, name: String = "L2Normalize"): Output

def linear(x: Output, weights: Output, bias: Output = null, name: String = "Linear"): Output

def localResponseNormalization(input: Output, depthRadius: Int = 5, bias: Float = 1.0f, alpha: Float = 1.0f, beta: Float = 0.5f, name: String = "LocalResponseNormalization"): Output

def logPoissonLoss(logPredictions: Output, targets: Output, computeFullLoss: Boolean = false, name: String = "LogPoissonLoss"): Output

def logSoftmax(logits: Output, axis: Int = 1, name: String = "LogSoftmax"): Output

def lrn(input: Output, depthRadius: Int = 5, bias: Float = 1.0f, alpha: Float = 1.0f, beta: Float = 0.5f, name: String = "LRN"): Output

def maxPool(input: Output, windowSize: Seq[Long], stride1: Long, stride2: Long, padding: ConvPaddingMode, dataFormat: CNNDataFormat = CNNDataFormat.default, name: String = "MaxPool"): Output

def maxPoolGrad(originalInput: Output, originalOutput: Output, outputGradient: Output, windowSize: Seq[Long], stride1: Long, stride2: Long, padding: ConvPaddingMode, dataFormat: CNNDataFormat = CNNDataFormat.default, name: String = "MaxPoolGrad"): Output

def maxPoolGradGrad(originalInput: Output, originalOutput: Output, outputGradient: Output, windowSize: Seq[Long], stride1: Long, stride2: Long, padding: ConvPaddingMode, dataFormat: CNNDataFormat = CNNDataFormat.default, name: String = "MaxPoolGradGrad"): Output

final def ne(arg0: AnyRef): Boolean

final def notify(): Unit

final def notifyAll(): Unit

def relu(input: Output, alpha: Float = 0.0f, name: String = "ReLU"): Output

def relu6[T](x: T, name: String = "ReLU6")(implicit arg0: OutputOps[T]): T

def selu[T](x: T, name: String = "SELU")(implicit arg0: OutputOps[T]): T

def sequenceLoss(logits: Output, labels: Output, weights: Output = null, averageAcrossTimeSteps: Boolean = true, averageAcrossBatch: Boolean = true, lossFn: (Output, Output) ⇒ Output = sparseSoftmaxCrossEntropy(_, _), name: String = "SequenceLoss"): Output

def sigmoidCrossEntropy(logits: Output, labels: Output, weights: Output = null, name: String = "SigmoidCrossEntropy"): Output

def softmax(logits: Output, axis: Int = 1, name: String = "Softmax"): Output

def softmaxCrossEntropy(logits: Output, labels: Output, axis: Int = 1, name: String = "SoftmaxCrossEntropy"): Output

def softplus[T](x: T, name: String = "Softplus")(implicit arg0: OutputOps[T]): T

def softsign[T](x: T, name: String = "Softsign")(implicit arg0: OutputOps[T]): T

def sparseSoftmaxCrossEntropy(logits: Output, labels: Output, axis: Int = 1, name: String = "SparseSoftmaxCrossEntropy"): Output

final def synchronized[T0](arg0: ⇒ T0): T0

def toString(): String

def topK(input: Output, k: Output = 1, sorted: Boolean = true, name: String = "TopK"): (Output, Output)

final def wait(): Unit

final def wait(arg0: Long, arg1: Int): Unit

final def wait(arg0: Long): Unit

Inherited from NN

Inherited from AnyRef

Inherited from Any

NNOps

Ungrouped