All Classes and Interfaces
Class
Description
Rectified Linear Unit (ReLU) activation function.
Implementation of
ActivationReLU for Tensor_F32.Implementation of
ActivationReLU for Tensor_F64.The sigmoid is defined as:
Implementation of
ActivationSigmoid for Tensor_F32.Implementation of
ActivationSigmoid for Tensor_F64.The hyperbolic tangent (tanh) is defined as:
Implementation of
ActivationTanH for Tensor_F32.Implementation of
ActivationTanH for Tensor_F64.Base class which implements common functionality between all
functionsAbstract class fo all virtual 2D spatial padding implementation.
Common class for implementations of
SpatialConvolve2D.Base class for all Tensor data types.
Batch Normalization [1] determines the mean and standard deviation (stdev) of each input element individually
using the training data.
Interface for padding in which the region being sampled has been clipped so that it will be
entirely contained inside the original image.
Implementation of
ConstantPadding2D_F32.Implementation of
ConstantPadding2D_F64.Configuration for 2D convolution.
Configuration for spatial padding.
Common configuration for many spatial functions
Complex algorithms with several parameters can specify their parameters using a separate class.
Interface for padding which applies a constant padding to the output of the image
Pads pixels outside the input image with a user specified constant value.
Pads pixels outside the input image with a user specified constant value.
Various constants used throughout DeepBoof
Automatically generated file containing build version information.
Functions which also implement the backwards step and compute the gradient for all inputs.Base class for element-wise functions
High level interface for functions in an Artificial Neural Network.
Implementation of a forward only Batch Normalization.
Implementation of
FunctionBatchNorm for Tensor_F32.Implementation of
FunctionBatchNorm for Tensor_F64.Multiplies each element in a tensor by the same value.
Implementation of
FunctionElementWiseMult for Tensor_F32.Implementation of
FunctionElementWiseMult for Tensor_F64.Applies a linear (or affine) equation to input array.
Implementation of
FunctionLinear for Tensor_F32.Implementation of
FunctionLinear for Tensor_F64.Processes a sequence of forward functions.
Address which points to the input of a node
Tensor Interface
Node in a network graph which describes the network's processing sequence.
Specifies the type of padding applied to a spacial function.
Orders an unsorted list of nodes so that they can be processed in sequence and have all of their dependencies meet
prior to being invoked.
Max spatial pooling find the average value inside the pooling region.
Implementation of
SpatialAveragePooling for Tensor_F32.Implementation of
SpatialAveragePooling for Tensor_F64.Spatial
Batch Normalization seeks to maintain the convolutional property, "that
different elements of the same feature map, at different locations, are normalized in the same way." [1]
Thus the input tensor (N,C,H,W) is "reshaped" such that it is (N*H*W,C) and it's treated like a mini-batch
with N*H*W elements.Implementation of
SpatialBatchNorm for Tensor_F32Implementation of
SpatialBatchNorm for Tensor_F64Performs convolutions across an input image with special kernels that have 'C' channels, one for each input image.
Implementation of
SpatialConvolve2D for Tensor_F32Implementation of
SpatialConvolve2D for Tensor_F64Max spatial pooling find the maximum value inside the pooling region.
Implementation of
SpatialMaxPooling for Tensor_F32.Implementation of
SpatialMaxPooling for Tensor_F64.Interface for all virtual 2D spatial padding implementation.
Abstract class for F64 implementations of
BaseSpatialPadding2D.Abstract class for F64 implementations of
BaseSpatialPadding2D.Spatial pooling down samples the input spatial tensors by finding a representative value inside
each pooling region.
Implementation of
BaseSpatialWindow which processes the spatial tensor is processed in
BCHW (mini-batch, channel, height, width) orderImplementation of
BaseSpatialWindow which processes the spatial tensor is one
image in a mini batch at a time.Base class for Tensors.
Various functions for unit tests
Various functions for unit tests
Merged multiple input tensors into a single output which can be processed by a
Function.A virtual tensor doesn't physically store in memory the entire tensor, instead it uses an equation
to generate on the fly some or all of a tensor.
Virtual tensor for 64bit float types.
Virtual tensor for 64bit float types.