Simd Library Documentation.

Home | Release Notes | Download | Documentation | Forum | SourceForge | GitHub
ConvolutionalLayer Class Reference

ConfolutionLayer class. More...

+ Inheritance diagram for ConvolutionalLayer:

Public Member Functions

 ConvolutionalLayer (Function::Type f, const Size &srcSize, size_t srcDepth, size_t dstDepth, size_t coreSize, bool valid=true, bool bias=true, const View &connection=View())
 Creates new ConfolutionLayer class. More...
 
- Public Member Functions inherited from Layer
virtual ~Layer ()
 

Additional Inherited Members

- Public Types inherited from Layer
enum  Type {
  Input,
  Convolutional,
  MaxPooling,
  FullyConnected,
  Dropout
}
 
enum  Method {
  Fast,
  Check,
  Train
}
 

Detailed Description

ConfolutionLayer class.

Convolutional layer in neural network.

Constructor & Destructor Documentation

ConvolutionalLayer ( Function::Type  f,
const Size srcSize,
size_t  srcDepth,
size_t  dstDepth,
size_t  coreSize,
bool  valid = true,
bool  bias = true,
const View connection = View() 
)

Creates new ConfolutionLayer class.

Parameters
[in]f- a type of activation function used in this layer.
[in]srcSize- a size (width and height) of input image.
[in]srcDepth- a number of input channels (images).
[in]dstDepth- a number of output channels (images).
[in]coreSize- a size of convolution core.
[in]valid- a boolean flag (True - only original image points are used in convolution, so output image is decreased; False - input image is padded by zeros and output image has the same size). By default its true.
[in]bias- a boolean flag (enabling of bias). By default its True.
[in]connection- a table of connections between input and output channels. By default all channels are connected.