dcnnasgn::loss_layer< CSP, PP > Struct Template Reference

The loss layer. More...

#include <dcnnelements.hpp>

Inheritance diagram for dcnnasgn::loss_layer< CSP, PP >:

Static Public Member Functions
static void	forward (const input_data &ind, const gold_data &gd, output_data &outd)
	The forward-propagation function of the loss layer.
Static Public Member Functions inherited from dcnnasgn::loss_layer_base< CSP, PP >
static tagged::range_class< batch_tag >	forward_check (const input_data &ind, const gold_data &gd, output_data &outd)

Additional Inherited Members
Public Types inherited from dcnnasgn::loss_layer_base< CSP, PP >
using	input_data = dcnnsol::feature_data<CSP, PP>
using	output_data = loss_data
using	channel_tag = typename CSP::channel_tag
Static Public Attributes inherited from dcnnasgn::loss_layer_base< CSP, PP >
static constexpr auto	cr = CSP::cr

Detailed Description

template<typename CSP, is_policy PP>
struct dcnnasgn::loss_layer< CSP, PP >

The loss layer.

The loss function is the sum of square differences between network output and ground truth.

Template Parameters

CSP	Channel size policy

Member Function Documentation

forward()

template<typename CSP, is_policy PP>

static void dcnnasgn::loss_layer< CSP, PP >::forward ( const input_data & ind, const gold_data & gd, output_data & outd )

inlinestatic

The forward-propagation function of the loss layer.

Parameters

ind	Input activations
gd	Ground-truth activations
outd	The loss

---

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

• asgn/dcnnelements.hpp