Package deepnetts.net

Class ConvolutionalNetwork.Builder

java.lang.Object

deepnetts.net.ConvolutionalNetwork.Builder

Enclosing class:

ConvolutionalNetwork

public static class ConvolutionalNetwork.Builder
extends Object

Builder for a convolutional neural network.

See Also:

• ConvolutionalNetwork

Constructor Summary

Constructors

Constructor	Description
Builder()

Method Summary

Modifier and Type	Method	Description
ConvolutionalNetwork.Builder	addConvolutionalLayer(int channelNum)	Adds a convolutional layer with the given number of channels(filters), with default 3x3 filter size and default activation function.
ConvolutionalNetwork.Builder	addConvolutionalLayer(int channelNum, int filterSize)	Adds a convolutional layer with the given number of channels(filters), with given filter size (same width and height) and default activation function.
ConvolutionalNetwork.Builder	addConvolutionalLayer(int channelNum, int filterWidth, int filterHeight)	Adds a convolutional layer with the given number of channels(filters), with given width and height of convolutional filter and default type of activation function.
ConvolutionalNetwork.Builder	addConvolutionalLayer(int channelNum, int filterWidth, int filterHeight, int stride)	Adds a convolutional layer with the given width and height of convolutional filters, given number of channels(filters) and stride(filter step), and default type of activation function.
ConvolutionalNetwork.Builder	addConvolutionalLayer(int channelNum, int filterWidth, int filterHeight, int stride, ActivationType activationType)	Adds a convolutional layer with the given width and height of convolutional filters, given number of channels(filters) and stride(filter step), and given type of activation function.
ConvolutionalNetwork.Builder	addConvolutionalLayer(int channelNum, int filterSize, ActivationType activationType)	Adds a convolutional layer with the given number of channels(filters), with given filter size (same width and height) and given type of activation function.
ConvolutionalNetwork.Builder	addConvolutionalLayer(int channelNum, ActivationType activationType)	Adds a convolutional layer with the given number of channels(filters), with given activation function type and default 3x3 filter size.
ConvolutionalNetwork.Builder	addConvolutionalLayer(int channelNum, Filter filter)	Adds a convolutional layer with the given number of channels(filters), with given settings of a convolutional filter and default type of activation function.
ConvolutionalNetwork.Builder	addConvolutionalLayer(int channelNum, Filter filter, ActivationType activationType)	Adds a convolutional layer with the given number of channels(filters), with given settings of a convolutional filter and given type of activation function.
ConvolutionalNetwork.Builder	addFullyConnectedLayer(int layerWidth)	Adds fully connected layer with specified width and default activation function.
ConvolutionalNetwork.Builder	addFullyConnectedLayer(int layerWidth, ActivationType activationType)	Adds fully connected layer with specified width and activation function.
ConvolutionalNetwork.Builder	addInputLayer(int width, int height)	Input layer with specified width and height, and 3 channels by default.
ConvolutionalNetwork.Builder	addInputLayer(int width, int height, int channels)	Input layer with specified width, height and number of channels (depth).
ConvolutionalNetwork.Builder	addLayer(AbstractLayer layer)	Adds a given layer to the network.
ConvolutionalNetwork.Builder	addMaxPoolingLayer(int filterSize)
ConvolutionalNetwork.Builder	addMaxPoolingLayer(int filterSize, int stride)	Adds a max pooling layer with given filter size and stride(filter step).
ConvolutionalNetwork.Builder	addMaxPoolingLayer(int filterWidth, int filterHeight, int stride)	Adds a max pooling layer with given filter size and stride(filter step).Max pooling layer comes after convolutional layer and reduces the dimensions of the input received from the previous layer.Typically filter sizes of 2 are used, which effectively halves the dimensions of the input from the previous layer.
ConvolutionalNetwork.Builder	addMaxPoolingLayer(Filter filter)	Adds a max pooling layer with the given filter settings.Max pooling layer comes after convolutional layer and reduces the dimensions of the input received from the previous layer.
ConvolutionalNetwork.Builder	addOutputLayer(int layerWidth, ActivationType activationType)	Adds output layer to the neural network with specified width (number of outputs) and activation function type.
ConvolutionalNetwork.Builder	addOutputLayer(int layerWidth, Class<? extends OutputLayer> clazz)	Adds output layer to the neural network with specified width (number of outputs) and layer class.
ConvolutionalNetwork	build()	Builds an instance of ConvolutionalNetwork with settings specified in this builder.
ConvolutionalNetwork.Builder	hiddenActivationFunction(ActivationType activationType)	Sets default type of the activation function to use for all hidden layers in the network.
ConvolutionalNetwork.Builder	lossFunction(LossType lossType)	Sets loss function to be used by created neural network.
ConvolutionalNetwork.Builder	randomSeed(long seed)	Initializes random number generator with the specified seed in order to get same random number sequences used for weights initialization.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Constructor Details

Builder

public Builder()

Method Details

addInputLayer

public ConvolutionalNetwork.Builder addInputLayer(int width,
 int height)

Input layer with specified width and height, and 3 channels by default.

Parameters:

width -

height -

Returns:

addInputLayer

public ConvolutionalNetwork.Builder addInputLayer(int width,
 int height,
 int channels)

Input layer with specified width, height and number of channels (depth).

Parameters:

width - width of the input Tensor

height - height of the input Tensor

channels - depth of the input Tensor

Returns:

this builder

addFullyConnectedLayer

public ConvolutionalNetwork.Builder addFullyConnectedLayer(int layerWidth)

Adds fully connected layer with specified width and default activation function. In fully connected layer each neuron is connected to all outputs from previous layer.

Parameters:

layerWidth - width of the layer which corresponds to the number of outputs/neurons in this layer

Returns:

current builder instance

See Also:

• FullyConnectedLayer
• ActivationType

addFullyConnectedLayer

public ConvolutionalNetwork.Builder addFullyConnectedLayer(int layerWidth,
 ActivationType activationType)

Adds fully connected layer with specified width and activation function. In dense layer each neuron is connected to all outputs from previous layer.

Parameters:

layerWidth - width of the layer which corresponds to the number of outputs/neurons in this layer

activationType - type of activation function

Returns:

current builder instance

See Also:

• FullyConnectedLayer
• ActivationType

addOutputLayer

public ConvolutionalNetwork.Builder addOutputLayer(int layerWidth,
 Class<? extends OutputLayer> clazz)

Adds output layer to the neural network with specified width (number of outputs) and layer class.

Parameters:

layerWidth - width of the layer which corresponds to number of network's outputs

clazz - class of the output layer

Returns:

builder instance

addOutputLayer

public ConvolutionalNetwork.Builder addOutputLayer(int layerWidth,
 ActivationType activationType)

Adds output layer to the neural network with specified width (number of outputs) and activation function type.

Parameters:

layerWidth - width of the layer which corresponds to number of network's outputs

activationType - type of the activation function to use in output layer

Returns:

builder instance

addConvolutionalLayer

public ConvolutionalNetwork.Builder addConvolutionalLayer(int channelNum)

Adds a convolutional layer with the given number of channels(filters), with default 3x3 filter size and default activation function. Each channel(filter) is capable to learn to detect a specific patern of pixels in image.

Parameters:

channelNum - number of channels(filters) in the convolutional layer

Returns:

builder instance

See Also:

• ConvolutionalLayer

addConvolutionalLayer

public ConvolutionalNetwork.Builder addConvolutionalLayer(int channelNum,
 ActivationType activationType)

Adds a convolutional layer with the given number of channels(filters), with given activation function type and default 3x3 filter size. Each channel(filter) is capable to learn to detect a specific patern of pixels in image.

Parameters:

channelNum - number of channels(filters) in the convolutional layer

activationType - type of the activation function in the convolutional layer

Returns:

builder instance

See Also:

• ConvolutionalLayer
• ActivationType

addConvolutionalLayer

public ConvolutionalNetwork.Builder addConvolutionalLayer(int channelNum,
 int filterSize)

Adds a convolutional layer with the given number of channels(filters), with given filter size (same width and height) and default activation function. Each channel(filter) is capable to learn to detect a specific patern of pixels in image.

Parameters:

channelNum - number of channels(filters) in the convolutional layer

filterSize - size of the convolutional filter (same width and height)

Returns:

builder instance

See Also:

• ConvolutionalLayer

addConvolutionalLayer

public ConvolutionalNetwork.Builder addConvolutionalLayer(int channelNum,
 int filterSize,
 ActivationType activationType)

Adds a convolutional layer with the given number of channels(filters), with given filter size (same width and height) and given type of activation function. Each channel(filter) is capable to learn to detect a specific patern of pixels in image.

Parameters:

channelNum - number of channels(filters) in the convolutional layer

filterSize - size of the convolutional filter (same width and height)

activationType - type of the activation function in the convolutional layer

Returns:

builder instance

See Also:

• ConvolutionalLayer
• ActivationType

addConvolutionalLayer

public ConvolutionalNetwork.Builder addConvolutionalLayer(int channelNum,
 int filterWidth,
 int filterHeight)

Adds a convolutional layer with the given number of channels(filters), with given width and height of convolutional filter and default type of activation function. Each channel(filter) is capable to learn to detect a specific patern of pixels in image.

Parameters:

channelNum - number of channels(filters) in the convolutional layer

filterWidth - width of a convolutional filter

filterHeight - height of a convolutional filter

Returns:

builder instance

See Also:

• ConvolutionalLayer

addConvolutionalLayer

public ConvolutionalNetwork.Builder addConvolutionalLayer(int channelNum,
 Filter filter)

Adds a convolutional layer with the given number of channels(filters), with given settings of a convolutional filter and default type of activation function. Each channel(filter) is capable to learn to detect a specific patern of pixels in image.

Parameters:

channelNum - number of channels(filters) in the convolutional layer

filter - settings of the convolutional filter(width, height, stride, padding)

Returns:

builder instance

See Also:

• ConvolutionalLayer

addConvolutionalLayer

public ConvolutionalNetwork.Builder addConvolutionalLayer(int channelNum,
 int filterWidth,
 int filterHeight,
 int stride)

Adds a convolutional layer with the given width and height of convolutional filters, given number of channels(filters) and stride(filter step), and default type of activation function. Each channel(filter) is capable to learn to detect a specific patern of pixels in image.

Parameters:

channelNum - number of channels(filters)

filterWidth - width of a convolutional filter

filterHeight - height of a convolutional filter

stride - filter stride(step)

Returns:

builder instance

See Also:

• ConvolutionalLayer

addConvolutionalLayer

public ConvolutionalNetwork.Builder addConvolutionalLayer(int channelNum,
 Filter filter,
 ActivationType activationType)

Adds a convolutional layer with the given number of channels(filters), with given settings of a convolutional filter and given type of activation function. Each channel(filter) is capable to learn to detect a specific patern of pixels in image.

Parameters:

channelNum - number of channels(filters) in the convolutional layer

filter - settings of the convolutional filter(width, height, stride, padding)

activationType - type of the activation function in the convolutional layer

Returns:

builder instance

See Also:

• ConvolutionalLayer
• ActivationType

addConvolutionalLayer

public ConvolutionalNetwork.Builder addConvolutionalLayer(int channelNum,
 int filterWidth,
 int filterHeight,
 int stride,
 ActivationType activationType)

Adds a convolutional layer with the given width and height of convolutional filters, given number of channels(filters) and stride(filter step), and given type of activation function. Each channel(filter) is capable to learn to detect a specific patern of pixels in image.

Parameters:

channelNum - number of channels(filters)

filterWidth - width of a convolutional filter

filterHeight - height of a convolutional filter

stride - filter stride(step)

activationType - type of the activation function in the convolutional layer

Returns:

builder instance

See Also:

• ConvolutionalLayer
• ActivationType

addMaxPoolingLayer

public ConvolutionalNetwork.Builder addMaxPoolingLayer(int filterSize,
 int stride)

Adds a max pooling layer with given filter size and stride(filter step). Max pooling layer comes after convolutional layer and reduces the dimensions of the input received from the previous layer. Typically filter sizes of 2 are used, which effectively halves the dimensions of the input from the previous layer.

Parameters:

filterSize - size of the max pooling filter(typically 2)

stride - filter step size - how many pixels will filter slide. Typically same as filter size for pooling layers.

Returns:

builder instance

addMaxPoolingLayer

public ConvolutionalNetwork.Builder addMaxPoolingLayer(int filterSize)

addMaxPoolingLayer

public ConvolutionalNetwork.Builder addMaxPoolingLayer(int filterWidth,
 int filterHeight,
 int stride)

Adds a max pooling layer with given filter size and stride(filter step).Max pooling layer comes after convolutional layer and reduces the dimensions of the input received from the previous layer.Typically filter sizes of 2 are used, which effectively halves the dimensions of the input from the previous layer.

Parameters:

filterWidth - pooling filter width

filterHeight - pooling filter height

stride - filter step size which is typically same as filter size for pooling layers.

Returns:

builder instance

addMaxPoolingLayer

public ConvolutionalNetwork.Builder addMaxPoolingLayer(Filter filter)

Adds a max pooling layer with the given filter settings.Max pooling layer comes after convolutional layer and reduces the dimensions of the input received from the previous layer. Typically filter sizes of 2 are used, which effectively halves the dimensions of the input from the previous layer.

Parameters:

filter - settings of the pooling filter(width, height, stride, padding)

Returns:

builder instance

addLayer

public ConvolutionalNetwork.Builder addLayer(AbstractLayer layer)

Adds a given layer to the network.

Parameters:

layer -

Returns:

builder instance

hiddenActivationFunction

public ConvolutionalNetwork.Builder hiddenActivationFunction(ActivationType activationType)

Sets default type of the activation function to use for all hidden layers in the network.

Parameters:

activationType - type of activation function

Returns:

instance of the current builder

See Also:

• ActivationType

lossFunction

public ConvolutionalNetwork.Builder lossFunction(LossType lossType)

Sets loss function to be used by created neural network. Loss function calculates the network's error during the training as a difference between actual and target output provided in training set.

Parameters:

lossType - type of a loss function

Returns:

instance of the current builder

randomSeed

public ConvolutionalNetwork.Builder randomSeed(long seed)

Initializes random number generator with the specified seed in order to get same random number sequences used for weights initialization. Specifying this value enables getting same/repeatable initialization.

Parameters:

seed -

Returns:

instance of the current builder

build

public ConvolutionalNetwork build()

Builds an instance of ConvolutionalNetwork with settings specified in this builder.

Returns:

an instance of the ConvolutionalNetwork created by this builder