cldnn::lrn Struct Reference

Local response normalization. More...

#include <lrn.hpp>

Inheritance diagram for cldnn::lrn:

Collaboration diagram for cldnn::lrn:

Public Types
typedef cldnn_lrn_desc	dto

Public Member Functions
	lrn (const primitive_id &id, const primitive_id &input, uint32_t size, float k, float alpha, float beta, cldnn_lrn_norm_region lrn_norm_region, const padding &output_padding=padding())
	Constructs LRN primitive. More...
	lrn (const dto *dto)
	Constructs a copy from C API cldnn_normalization_desc.
Public Member Functions inherited from cldnn::primitive_base< lrn, cldnn_lrn_desc >
const cldnn_primitive_desc *	get_dto () const override
	Returns pointer to a C API primitive descriptor casted to cldnn_primitive_desc.
Public Member Functions inherited from cldnn::primitive
	primitive (const primitive_type_id &type, const primitive_id &id, const std::vector< primitive_id > &input, const padding &output_padding=padding())
	primitive (const cldnn_primitive_desc *dto)
	Constructs a copy from basic C API cldnn_primitive_desc.
std::vector< std::reference_wrapper< primitive_id > >	dependecies ()
	Returns references to all primitive ids on which this primitive depends - inputs, weights, biases, etc.
std::vector< primitive_id >	dependecies () const
	Returns copy of all primitive ids on which this primitive depends - inputs, weights, biases, etc.
	operator primitive_id () const
	Implicit conversion to primiitive id.

Static Public Member Functions
static primitive_type_id	type_id ()

Public Attributes
uint32_t	size
	Size of normalization.
float	k
	Hyper parameter "k".
float	alpha
	Hyper parameter "alpha".
float	beta
	Hyper parameter "beta".
cldnn_lrn_norm_region	norm_region
	Normalize across or within channel.
Public Attributes inherited from cldnn::primitive
const primitive_type_id	type
	Primitive's type id.
const primitive_id	id
	Primitive's id.
fixed_size_vector_ref	input
	List of ids of input primitives.
padding	output_padding
	Requested output padding.

Protected Member Functions
void	update_dto (dto &dto) const override
Protected Member Functions inherited from cldnn::primitive_base< lrn, cldnn_lrn_desc >
	primitive_base (const primitive_id &id, const std::vector< primitive_id > &input, const padding &output_padding=padding())
	primitive_base (const cldnn_lrn_desc *dto)
Protected Member Functions inherited from cldnn::primitive
virtual std::vector< std::reference_wrapper< const primitive_id > >	get_dependencies () const

Additional Inherited Members
Protected Attributes inherited from cldnn::primitive
primitive_id_arr	_input

Detailed Description

Local response normalization.

LRN layer as described in chapter 3.3 of "ImageNet Classification with Deep Convolutional Neural Networks" by Khrizevsky, Sutskever, Hinton.
See: http://www.cs.toronto.edu/~fritz/absps/imagenet.pdf

Alogrithm:

b(i,x,y) = a(i,x,y) / (k+alpha*sum(min(N-1, i+n/2); j=max(0,i-n/2); a(j,x,y)^2))

Where:

• b(i,x,y) : value at x, y from i-th feature map after normalization
• a(i,x,y) : value at x, y from i-th feature map before normalization
• N : number of feature maps
• n : size of normalization
• k, alpha, beta : hyper parameters (equal to 2, 10e-4, 0.75 in paper).

Definition at line 42 of file lrn.hpp.

Constructor & Destructor Documentation

lrn()

cldnn::lrn::lrn ( const primitive_id &  id, const primitive_id &  input, uint32_t  size, float  k, float  alpha, float  beta, cldnn_lrn_norm_region  lrn_norm_region, const padding &  output_padding = padding()  )

inline

Constructs LRN primitive.

Parameters

id	This primitive id.
input	Input primitive id.
size	Size of normalization.
k	Hyper parameter "k".
alpha	Hyper parameter "alpha".
beta	Hyper parameter "beta".
lrn_norm_region	Normalize across or within channel

Definition at line 54 of file lrn.hpp.

---

The documentation for this struct was generated from the following file:

• C:/ML/OpenSource/clDNN/api/CPP/lrn.hpp