Layers
Supported neural network layers.

What are layers?

Each layer in a neural network configuration represents a unit of hidden units. When layers are stacked together, they represent a deep neural network.

Using layers

All layers available in Eclipse Deeplearning4j can be used either in a MultiLayerNetwork or ComputationGraph. When configuring a neural network, you pass the layer configuration and the network will instantiate the layer for you.

Layers vs. vertices

If you are configuring complex networks such as InceptionV4, you will need to use the ComputationGraph API and join different branches together using vertices. Check the vertices for more information.

General layers

ActivationLayer

[source]
Activation layer is a simple layer that applies the specified activation function to the input activations
clone
1
public ActivationLayer clone()
Copied!
    param activation Activation function for the layer
activation
1
public Builder activation(String activationFunction)
Copied!
Activation function for the layer
activation
1
public Builder activation(IActivation activationFunction)
Copied!
    param activationFunction Activation function for the layer
activation
1
public Builder activation(Activation activation)
Copied!
    param activation Activation function for the layer

DenseLayer

[source]
Dense layer: a standard fully connected feed forward layer
hasBias
1
public Builder hasBias(boolean hasBias)
Copied!
If true (default): include bias parameters in the model. False: no bias.
hasLayerNorm
1
public Builder hasLayerNorm(boolean hasLayerNorm)
Copied!
If true (default = false): enable layer normalization on this layer

DropoutLayer

[source]
Dropout layer. This layer simply applies dropout at training time, and passes activations through unmodified at test
build
1
public DropoutLayer build()
Copied!
Create a dropout layer with standard {- link Dropout}, with the specified probability of retaining the input activation. See {- link Dropout} for the full details
    param dropout Activation retain probability.

EmbeddingLayer

[source]
Embedding layer: feed-forward layer that expects single integers per example as input (class numbers, in range 0 to the equivalent one-hot representation. Mathematically, EmbeddingLayer is equivalent to using a DenseLayer with a one-hot representation for the input; however, it can be much more efficient with a large number of classes (as a dense layer + one-hot input does a matrix multiply with all but one value being zero). Note: can only be used as the first layer for a network Note 2: For a given example index i, the output is activationFunction(weights.getRow(i) + bias), hence the weight rows can be considered a vector/embedding for each example. Note also that embedding layer has an activation function (set to IDENTITY to disable) and optional bias (which is disabled by default)
hasBias
1
public Builder hasBias(boolean hasBias)
Copied!
If true: include bias parameters in the layer. False (default): no bias.
weightInit
1
public Builder weightInit(EmbeddingInitializer embeddingInitializer)
Copied!
Initialize the embedding layer using the specified EmbeddingInitializer - such as a Word2Vec instance
    param embeddingInitializer Source of the embedding layer weights
weightInit
1
public Builder weightInit(INDArray vectors)
Copied!
Initialize the embedding layer using values from the specified array. Note that the array should have shape [vocabSize, vectorSize]. After copying values from the array to initialize the network parameters, the input array will be discarded (so that, if necessary, it can be garbage collected)
    param vectors Vectors to initialize the embedding layer with

EmbeddingSequenceLayer

[source]
Embedding layer for sequences: feed-forward layer that expects fixed-length number (inputLength) of integers/indices per example as input, ranged from 0 to numClasses - 1. This input thus has shape [numExamples, inputLength] or shape [numExamples, 1, inputLength]. The output of this layer is 3D (sequence/time series), namely of shape [numExamples, nOut, inputLength]. Note: can only be used as the first layer for a network Note 2: For a given example index i, the output is activationFunction(weights.getRow(i) + bias), hence the weight rows can be considered a vector/embedding of each index. Note also that embedding layer has an activation function (set to IDENTITY to disable) and optional bias (which is disabled by default)
hasBias
1
public Builder hasBias(boolean hasBias)
Copied!
If true: include bias parameters in the layer. False (default): no bias.
inputLength
1
public Builder inputLength(int inputLength)
Copied!
Set input sequence length for this embedding layer.
    param inputLength input sequence length
    return Builder
inferInputLength
1
public Builder inferInputLength(boolean inferInputLength)
Copied!
Set input sequence inference mode for embedding layer.
    param inferInputLength whether to infer input length
    return Builder
weightInit
1
public Builder weightInit(EmbeddingInitializer embeddingInitializer)
Copied!
Initialize the embedding layer using the specified EmbeddingInitializer - such as a Word2Vec instance
    param embeddingInitializer Source of the embedding layer weights
weightInit
1
public Builder weightInit(INDArray vectors)
Copied!
Initialize the embedding layer using values from the specified array. Note that the array should have shape [vocabSize, vectorSize]. After copying values from the array to initialize the network parameters, the input array will be discarded (so that, if necessary, it can be garbage collected)
    param vectors Vectors to initialize the embedding layer with

GlobalPoolingLayer

[source]
Global pooling layer - used to do pooling over time for RNNs, and 2d pooling for CNNs. Supports the following
Global pooling layer can also handle mask arrays when dealing with variable length inputs. Mask arrays are assumed to be 2d, and are fed forward through the network during training or post-training forward pass:
    Time series: mask arrays are shape [miniBatchSize, maxTimeSeriesLength] and contain values 0 or 1 only
    CNNs: mask have shape [miniBatchSize, height] or [miniBatchSize, width]. Important: the current implementation assumes that for CNNs + variable length (masking), the input shape is [miniBatchSize, channels, height, 1] or [miniBatchSize, channels, 1, width] respectively. This is the case with global pooling in architectures like CNN for sentence classification.
Behaviour with default settings:
    3d (time series) input with shape [miniBatchSize, vectorSize, timeSeriesLength] -> 2d output [miniBatchSize, vectorSize]
    4d (CNN) input with shape [miniBatchSize, channels, height, width] -> 2d output [miniBatchSize, channels]
    5d (CNN3D) input with shape [miniBatchSize, channels, depth, height, width] -> 2d output [miniBatchSize, channels]
Alternatively, by setting collapseDimensions = false in the configuration, it is possible to retain the reduced dimensions as 1s: this gives
    [miniBatchSize, vectorSize, 1] for RNN output,
    [miniBatchSize, channels, 1, 1] for CNN output, and
    [miniBatchSize, channels, 1, 1, 1] for CNN3D output.
poolingDimensions
1
public Builder poolingDimensions(int... poolingDimensions)
Copied!
Pooling type for global pooling
poolingType
1
public Builder poolingType(PoolingType poolingType)
Copied!
    param poolingType Pooling type for global pooling
collapseDimensions
1
public Builder collapseDimensions(boolean collapseDimensions)
Copied!
Whether to collapse dimensions when pooling or not. Usually you do want to do this. Default: true. If true:
    3d (time series) input with shape [miniBatchSize, vectorSize, timeSeriesLength] -> 2d output [miniBatchSize, vectorSize]
    4d (CNN) input with shape [miniBatchSize, channels, height, width] -> 2d output [miniBatchSize, channels]
    5d (CNN3D) input with shape [miniBatchSize, channels, depth, height, width] -> 2d output [miniBatchSize, channels]
If false:
    3d (time series) input with shape [miniBatchSize, vectorSize, timeSeriesLength] -> 3d output [miniBatchSize, vectorSize, 1]
    4d (CNN) input with shape [miniBatchSize, channels, height, width] -> 2d output [miniBatchSize, channels, 1, 1]
    5d (CNN3D) input with shape [miniBatchSize, channels, depth, height, width] -> 2d output [miniBatchSize, channels, 1, 1, 1]
    param collapseDimensions Whether to collapse the dimensions or not
pnorm
1
public Builder pnorm(int pnorm)
Copied!
P-norm constant. Only used if using {- link PoolingType#PNORM} for the pooling type
    param pnorm P-norm constant

LocalResponseNormalization

[source]
Local response normalization layer See section 3.3 of http://www.cs.toronto.edu/~fritz/absps/imagenet.pdf
k
1
public Builder k(double k)
Copied!
LRN scaling constant k. Default: 2
n
1
public Builder n(double n)
Copied!
Number of adjacent kernel maps to use when doing LRN. default: 5
    param n Number of adjacent kernel maps
alpha
1
public Builder alpha(double alpha)
Copied!
LRN scaling constant alpha. Default: 1e-4
    param alpha Scaling constant
beta
1
public Builder beta(double beta)
Copied!
Scaling constant beta. Default: 0.75
    param beta Scaling constant
cudnnAllowFallback
1
public Builder cudnnAllowFallback(boolean allowFallback)
Copied!
When using CuDNN and an error is encountered, should fallback to the non-CuDNN implementatation be allowed? If set to false, an exception in CuDNN will be propagated back to the user. If false, the built-in (non-CuDNN) implementation for BatchNormalization will be used
    param allowFallback Whether fallback to non-CuDNN implementation should be used

LocallyConnected1D

[source]
SameDiff version of a 1D locally connected layer.
nIn
1
public Builder nIn(int nIn)
Copied!
Number of inputs to the layer (input size)
nOut
1
public Builder nOut(int nOut)
Copied!
    param nOut Number of outputs (output size)
activation
1
public Builder activation(Activation activation)
Copied!
    param activation Activation function for the layer
kernelSize
1
public Builder kernelSize(int k)
Copied!
    param k Kernel size for the layer
stride
1
public Builder stride(int s)
Copied!
    param s Stride for the layer
padding
1
public Builder padding(int p)
Copied!
    param p Padding for the layer. Not used if {- link ConvolutionMode#Same} is set
convolutionMode
1
public Builder convolutionMode(ConvolutionMode cm)
Copied!
    param cm Convolution mode for the layer. See {- link ConvolutionMode} for details
dilation
1
public Builder dilation(int d)
Copied!
    param d Dilation for the layer
hasBias
1
public Builder hasBias(boolean hasBias)
Copied!
    param hasBias If true (default is false) the layer will have a bias
setInputSize
1
public Builder setInputSize(int inputSize)
Copied!
Set input filter size for this locally connected 1D layer
    param inputSize height of the input filters
    return Builder

LocallyConnected2D

[source]
SameDiff version of a 2D locally connected layer.
setKernel
1
public void setKernel(int... kernel)
Copied!
Number of inputs to the layer (input size)
setStride
1
public void setStride(int... stride)
Copied!
    param stride Stride for the layer. Must be 2 values (height/width)
setPadding
1
public void setPadding(int... padding)
Copied!
    param padding Padding for the layer. Not used if {- link ConvolutionMode#Same} is set. Must be 2 values (height/width)
setDilation
1
public void setDilation(int... dilation)
Copied!
    param dilation Dilation for the layer. Must be 2 values (height/width)
nIn
1
public Builder nIn(int nIn)
Copied!
    param nIn Number of inputs to the layer (input size)
nOut
1
public Builder nOut(int nOut)
Copied!
    param nOut Number of outputs (output size)
activation
1
public Builder activation(Activation activation)
Copied!
    param activation Activation function for the layer
kernelSize
1
public Builder kernelSize(int... k)
Copied!
    param k Kernel size for the layer. Must be 2 values (height/width)
stride
1
public Builder stride(int... s)
Copied!
    param s Stride for the layer. Must be 2 values (height/width)
padding
1
public Builder padding(int... p)
Copied!
    param p Padding for the layer. Not used if {- link ConvolutionMode#Same} is set. Must be 2 values (height/width)
convolutionMode
1
public Builder convolutionMode(ConvolutionMode cm)
Copied!
    param cm Convolution mode for the layer. See {- link ConvolutionMode} for details
dilation
1
public Builder dilation(int... d)
Copied!
    param d Dilation for the layer. Must be 2 values (height/width)
hasBias
1
public Builder hasBias(boolean hasBias)
Copied!
    param hasBias If true (default is false) the layer will have a bias
setInputSize
1
public Builder setInputSize(int... inputSize)
Copied!
Set input filter size (h,w) for this locally connected 2D layer
    param inputSize pair of height and width of the input filters to this layer
    return Builder

LossLayer

[source]
LossLayer is a flexible output layer that performs a loss function on an input without MLP logic. LossLayer is does not have any parameters. Consequently, setting nIn/nOut isn’t supported - the output size is the same size as the input activations.
nIn
1
public Builder nIn(int nIn)
Copied!
    param lossFunction Loss function for the loss layer

OutputLayer

[source]
Output layer used for training via backpropagation based on labels and a specified loss function. Can be configured for both classification and regression. Note that OutputLayer has parameters - it contains a fully-connected layer (effectively contains a DenseLayer) internally. This allows the output size to be different to the layer input size.
build
1
public OutputLayer build()
Copied!
    param lossFunction Loss function for the output layer

Pooling1D

[source]
Supports the following pooling types: MAX, AVG, SUM, PNORM, NONE

Pooling2D

[source]
Supports the following pooling types: MAX, AVG, SUM, PNORM, NONE

Subsampling1DLayer

[source]
sequenceLength]}. This layer accepts RNN InputTypes instead of CNN InputTypes.
Supports the following pooling types: MAX, AVG, SUM, PNORM
setKernelSize
1
public void setKernelSize(int... kernelSize)
Copied!
Kernel size
    param kernelSize kernel size
setStride
1
public void setStride(int... stride)
Copied!
Stride
    param stride stride value
setPadding
1
public void setPadding(int... padding)
Copied!
Padding
    param padding padding value

Upsampling1D

[source]
sequenceLength]} Example:
1
If input (for a single example, with channels down page, and sequence from left to right) is:
2
[ A1, A2, A3]
3
[ B1, B2, B3]
4
Then output with size = 2 is:
5
[ A1, A1, A2, A2, A3, A3]
6
[ B1, B1, B2, B2, B3, B2]
Copied!
size
1
public Builder size(int size)
Copied!
Upsampling size
    param size upsampling size in single spatial dimension of this 1D layer
size
1
public Builder size(int[] size)
Copied!
Upsampling size int array with a single element. Array must be length 1
    param size upsampling size in single spatial dimension of this 1D layer

Upsampling2D

[source]
Upsampling 2D layer Repeats each value (or rather, set of depth values) in the height and width dimensions by
1
Input (slice for one example and channel)
2
[ A, B ]
3
[ C, D ]
4
Size = [2, 2]
5
Output (slice for one example and channel)
6
[ A, A, B, B ]
7
[ A, A, B, B ]
8
[ C, C, D, D ]
9
[ C, C, D, D ]
Copied!
size
1
public Builder size(int size)
Copied!
Upsampling size int, used for both height and width
    param size upsampling size in height and width dimensions
size
1
public Builder size(int[] size)
Copied!
Upsampling size array
    param size upsampling size in height and width dimensions

Upsampling3D

[source]
Upsampling 3D layer Repeats each value (all channel values for each x/y/z location) by size[0], size[1] and [minibatch, channels, size[0] depth, size[1] height, size[2] width]}
size
1
public Builder size(int size)
Copied!
Upsampling size as int, so same upsampling size is used for depth, width and height
    param size upsampling size in height, width and depth dimensions
size
1
public Builder size(int[] size)
Copied!
Upsampling size as int, so same upsampling size is used for depth, width and height
    param size upsampling size in height, width and depth dimensions

ZeroPadding1DLayer

[source]
Zero padding 1D layer for convolutional neural networks. Allows padding to be done separately for top and bottom.
setPadding
1
public void setPadding(int... padding)
Copied!
Padding value for left and right. Must be length 2 array
build
1
public ZeroPadding1DLayer build()
Copied!
    param padding Padding for both the left and right

ZeroPadding3DLayer

[source]
Zero padding 3D layer for convolutional neural networks. Allows padding to be done separately for “left” and “right” in all three spatial dimensions.
setPadding
1
public void setPadding(int... padding)
Copied!
[padLeftD, padRightD, padLeftH, padRightH, padLeftW, padRightW]
build
1
public ZeroPadding3DLayer build()
Copied!
    param padding Padding for both the left and right in all three spatial dimensions

ZeroPaddingLayer

[source]
Zero padding layer for convolutional neural networks (2D CNNs). Allows padding to be done separately for top/bottom/left/right
setPadding
1
public void setPadding(int... padding)
Copied!
Padding value for top, bottom, left, and right. Must be length 4 array
build
1
public ZeroPaddingLayer build()
Copied!
    param padHeight Padding for both the top and bottom
    param padWidth Padding for both the left and right

ElementWiseMultiplicationLayer

[source]
is a learnable weight vector of length nOut
    “.” is element-wise multiplication
    b is a bias vector
Note that the input and output sizes of the element-wise layer are the same for this layer
created by jingshu
getMemoryReport
1
public LayerMemoryReport getMemoryReport(InputType inputType)
Copied!
This is a report of the estimated memory consumption for the given layer
    param inputType Input type to the layer. Memory consumption is often a function of the input type
    return Memory report for the layer

RepeatVector

[source]
RepeatVector layer configuration.
RepeatVector takes a mini-batch of vectors of shape (mb, length) and a repeat factor n and outputs a 3D tensor of shape (mb, n, length) in which x is repeated n times.
getRepetitionFactor
1
public int getRepetitionFactor()
Copied!
Set repetition factor for RepeatVector layer
setRepetitionFactor
1
public void setRepetitionFactor(int n)
Copied!
Set repetition factor for RepeatVector layer
    param n upsampling size in height and width dimensions
repetitionFactor
1
public Builder repetitionFactor(int n)
Copied!
Set repetition factor for RepeatVector layer
    param n upsampling size in height and width dimensions

Yolo2OutputLayer

[source]
Output (loss) layer for YOLOv2 object detection model, based on the papers: YOLO9000: Better, Faster, Stronger - Redmon & Farhadi (2016) - https://arxiv.org/abs/1612.08242 and You Only Look Once: Unified, Real-Time Object Detection - Redmon et al. (2016) - http://www.cv-foundation.org/openaccess/content_cvpr_2016/papers/Redmon_You_Only_Look_CVPR_2016_paper.pdf This loss function implementation is based on the YOLOv2 version of the paper. However, note that it doesn’t currently support simultaneous training on both detection and classification datasets as described in the YOlO9000 paper.
Note: Input activations to the Yolo2OutputLayer should have shape: [minibatch, b(5+c), H, W], where: b = number of bounding boxes (determined by config - see papers for details) c = number of classes H = output/label height W = output/label width
Important: In practice, this means that the last convolutional layer before your Yolo2OutputLayer should have output depth of b(5+c). Thus if you change the number of bounding boxes, or change the number of object classes, the number of channels (nOut of the last convolution layer) needs to also change. Label format: [minibatch, 4+C, H, W] Order for labels depth: [x1,y1,x2,y2,(class labels)] x1 = box top left position y1 = as above, y axis x2 = box bottom right position y2 = as above y axis Note: labels are represented as a multiple of grid size - for a 13x13 grid, (0,0) is top left, (13,13) is bottom right Note also that mask arrays are not required - this implementation infers the presence or absence of objects in each grid cell from the class labels (which should be 1-hot if an object is present, or all 0s otherwise).
lambdaCoord
1
public Builder lambdaCoord(double lambdaCoord)
Copied!
Loss function coefficient for position and size/scale components of the loss function. Default (as per paper): 5
lambbaNoObj
1
public Builder lambbaNoObj(double lambdaNoObj)
Copied!
Loss function coefficient for the “no object confidence” components of the loss function. Default (as per paper): 0.5
    param lambdaNoObj Lambda value for no-object (confidence) component of the loss function
lossPositionScale
1
public Builder lossPositionScale(ILossFunction lossPositionScale)
Copied!
Loss function for position/scale component of the loss function
    param lossPositionScale Loss function for position/scale
lossClassPredictions
1
public Builder lossClassPredictions(ILossFunction lossClassPredictions)
Copied!
Loss function for the class predictions - defaults to L2 loss (i.e., sum of squared errors, as per the paper), however Loss MCXENT could also be used (which is more common for classification).
    param lossClassPredictions Loss function for the class prediction error component of the YOLO loss function
boundingBoxPriors
1
public Builder boundingBoxPriors(INDArray boundingBoxes)
Copied!
Bounding box priors dimensions [width, height]. For N bounding boxes, input has shape [rows, columns] = [N, 2] Note that dimensions should be specified as fraction of grid size. For example, a network with 13x13 output, a value of 1.0 would correspond to one grid cell; a value of 13 would correspond to the entire image.
    param boundingBoxes Bounding box prior dimensions (width, height)

MaskLayer

[source]
MaskLayer applies the mask array to the forward pass activations, and backward pass gradients, passing through this layer. It can be used with 2d (feed-forward), 3d (time series) or 4d (CNN) activations.

MaskZeroLayer

[source]
Wrapper which masks timesteps with activation equal to the specified masking value (0.0 default). Assumes that the input shape is [batch_size, input_size, timesteps].