BatchNorm

class BatchNorm(axis: List<Int>, momentum: Double, center: Boolean, epsilon: Double, scale: Boolean, gammaInitializer: Initializer, betaInitializer: Initializer, gammaRegularizer: Regularizer?, betaRegularizer: Regularizer?, movingMeanInitializer: Initializer, movingVarianceInitializer: Initializer, name: String) : Layer, NoGradients, ParametrizedLayer

NOTE: This layer is not trainable and does not update its weights. It's frozen by default.

Since

0.2

Constructors

BatchNorm

fun BatchNorm(axis: List<Int> = arrayListOf(3), momentum: Double = 0.99, center: Boolean = true, epsilon: Double = 0.001, scale: Boolean = true, gammaInitializer: Initializer = Ones(), betaInitializer: Initializer = Zeros(), gammaRegularizer: Regularizer? = null, betaRegularizer: Regularizer? = null, movingMeanInitializer: Initializer = Zeros(), movingVarianceInitializer: Initializer = Ones(), name: String = "")

Creates BatchNorm object.

Functions

build

open override fun build(tf: Ops, inputShape: Shape)

Extend this function to define variables in layer.

buildFromInboundLayers

fun buildFromInboundLayers(tf: Ops)

Extend this function to define variables in layer.

computeOutputShape

open override fun computeOutputShape(inputShape: Shape): Shape

Computes output shape, based on inputShape and Layer type.

computeOutputShapeFromInboundLayers

open fun computeOutputShapeFromInboundLayers(): TensorShape

Computes output shape, based on input shapes of inbound layers.

forward

open override fun forward(tf: Ops, input: Operand<Float>, isTraining: Operand<Boolean>, numberOfLosses: Operand<Float>?): Operand<Float>

Builds main layer input transformation with tf. Depends on Layer type.

open fun forward(tf: Ops, input: List<Operand<Float>>, isTraining: Operand<Boolean>, numberOfLosses: Operand<Float>?): Operand<Float>

Builds main layer input transformation with tf. Depends on Layer type.

invoke

operator fun invoke(vararg layers: Layer): Layer

Important part of functional API. It takes layers as input and saves them to the inboundLayers of the given layer.

toString

open override fun toString(): String

Properties

axis

val axis: List<Int>

Integer or a list of integers, the axis that should be normalized (typically the features' axis).

betaInitializer

val betaInitializer: Initializer

Initializer for the beta weight.

betaRegularizer

val betaRegularizer: Regularizer? = null

Optional regularizer for the beta weight.

center

val center: Boolean = true

If True, add offset of beta to normalized tensor. If False, beta is ignored.

epsilon

val epsilon: Double = 0.001

Small float added to variance to avoid dividing by zero.

gammaInitializer

val gammaInitializer: Initializer

Initializer for the gamma weight.

gammaRegularizer

val gammaRegularizer: Regularizer? = null

Optional regularizer for the gamma weight.

hasActivation

open override val hasActivation: Boolean

Returns True, if layer has internal activation function.

inboundLayers

var inboundLayers: MutableList<Layer>

Returns inbound layers.

momentum

val momentum: Double = 0.99

Momentum for the moving average.

movingMeanInitializer

val movingMeanInitializer: Initializer

Initializer for the moving mean.

movingVarianceInitializer

val movingVarianceInitializer: Initializer

Initializer for the moving variance.

name

var name: String

Layer name. A new name is generated during model compilation when provided name is empty.

outboundLayers

var outboundLayers: MutableList<Layer>

Returns outbound layers.

outputShape

lateinit var outputShape: TensorShape

Output data tensor shape.

paramCount

open val paramCount: Int

Number of parameters in this layer.

parentModel

var parentModel: GraphTrainableModel? = null

Model where this layer is used.

scale

val scale: Boolean = true

If True, multiply by gamma. If False, gamma is not used.

variables

open override val variables: List<KVariable>

Variables used in this layer.