cldnn::activation Struct Reference

Activation using rectified linear unit or parameterized rectified linear unit. More...

#include <activation.hpp>

Inheritance diagram for cldnn::activation:

Collaboration diagram for cldnn::activation:

Public Types
typedef cldnn_activation_desc	dto

Public Member Functions
	activation (const primitive_id &id, const primitive_id &input, cldnn_activation_func activation_func, cldnn_activation_additional_params additional_params={ 0.f, 0.f }, const padding &output_padding=padding())
	Constructs Relu primitive. More...
	activation (const primitive_id &id, const primitive_id &input, const primitive_id &additional_params_input, cldnn_activation_func activation_func, const padding &output_padding=padding())
	Constructs activation with input per feature. More...
	activation (const dto *dto)
	Constructs a copy from basic C API cldnn_activation_desc.
Public Member Functions inherited from cldnn::primitive_base< activation, cldnn_activation_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
cldnn_activation_func	activation_func
	activation function.
cldnn_activation_additional_params	additional_params
	activation additional params.
primitive_id	additional_params_input
	PRelu activation slope input primitive id. Input x dimension should be equal to input feature size (one slope per channel). All other dimensions should be 1.
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
std::vector< std::reference_wrapper< const primitive_id > >	get_dependencies () const override
void	update_dto (dto &dto) const override
Protected Member Functions inherited from cldnn::primitive_base< activation, cldnn_activation_desc >
	primitive_base (const primitive_id &id, const std::vector< primitive_id > &input, const padding &output_padding=padding())
	primitive_base (const cldnn_activation_desc *dto)

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

Detailed Description

Activation using rectified linear unit or parameterized rectified linear unit.

Can get one negative slope or negative slope per channel.

Algorithm:

out(i,x,y) = max(0, in(i,x,y)) + slope(i) * min(0, in(i,x,y))

Where:

• out(i,x,y) : value at x, y from i-th feature map after activation.
• in(i,x,y) : value at x, y from i-th feature map before activation.
• slope(i) : the slope value of the i-th feature map (can be shared across channels or one slope per channel).

Definition at line 39 of file activation.hpp.

Constructor & Destructor Documentation

activation() [1/2]

cldnn::activation::activation ( const primitive_id &  id, const primitive_id &  input, cldnn_activation_func  activation_func, cldnn_activation_additional_params  additional_params = { 0.f,0.f }, const padding &  output_padding = padding()  )

inline

Constructs Relu primitive.

Parameters

id	This primitive id.
input	Input primitive id.
activation_func	activation function.
additional_params	additional params (slope/max_val/linear a,b).

Definition at line 48 of file activation.hpp.

activation() [2/2]

cldnn::activation::activation ( const primitive_id &  id, const primitive_id &  input, const primitive_id &  additional_params_input, cldnn_activation_func  activation_func, const padding &  output_padding = padding()  )

inline

Constructs activation with input per feature.

Parameters

id	This primitive id.
input	Input primitive id.
additional_params_input	additional params stored on a memory. Input x dimension should be equal to input feature size (one value per channel. in case of linear is one pair per channel). All other dimensions should be 1.

Definition at line 68 of file activation.hpp.

---

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

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