asgn
 
Loading...
Searching...
No Matches
The complete network

Classes

struct  dcnnasgn::combined_policy< PP >
 Policy: The complete network. More...
 
class  dcnnasgn::combined_model< PP >
 Model data (weights and biases) of the complete network. More...
 
class  dcnnasgn::combined_data< PP >
 Input data, forward-propagated activations, and loss of the complete network. More...
 

Typedefs

using dcnnasgn::test_labels_t = tagged::tensor_class<gold_labels_policy::label_carrier, input_tag>
 
using dcnnasgn::test_images_t = tagged::tensor_class<idp::image_carrier, input_tag, idp::height_tag, idp::width_tag, idp::channel_tag>
 

Functions

template<is_policy PP>
void dcnnasgn::combined_load_model (combined_model< PP > &m, const std::filesystem::path &data_folder)
 
template<typename mapping, is_policy PP>
float dcnnasgn::combined_forward (const test_images_t &test_images, const test_labels_t &test_labels, mapping &&bmap, const combined_model< PP > &m, combined_data< PP > &d)
 The forward-propagation function of the complete network.
 

Detailed Description

Function Documentation

◆ combined_forward()

template<typename mapping, is_policy PP>
float dcnnasgn::combined_forward ( const test_images_t & test_images,
const test_labels_t & test_labels,
mapping && bmap,
const combined_model< PP > & m,
combined_data< PP > & d )
inline

The forward-propagation function of the complete network.

Includes creation of a minibatch by selection of the specified input images from the training dataset

Template Parameters
mappingInput selection mapping
PPPermutation policy
Parameters
train_imagesTraining image dataset
train_labelsGround-truth dataset
bmapInput selection mapping
mNetwork model data
dForward-propagation data
Returns
Total loss over the minibatch