XCSF  1.4.7
XCSF learning classifier system
Data Fields
ArgsLayer Struct Reference

Parameters for initialising a neural network layer. More...

#include <neural_layer_args.h>

Collaboration diagram for ArgsLayer:
[legend]

Data Fields

int type
 Layer type: CONNECTED, DROPOUT, etc. More...
 
int n_inputs
 Number of inputs. More...
 
int n_init
 Initial number of units / neurons / filters. More...
 
int n_max
 Maximum number of units / neurons. More...
 
int max_neuron_grow
 Maximum number neurons to add per mutation event. More...
 
int function
 Activation function. More...
 
int recurrent_function
 Recurrent activation function. More...
 
int height
 Pool, Conv, and Upsample. More...
 
int width
 Pool, Conv, and Upsample. More...
 
int channels
 Pool, Conv, and Upsample. More...
 
int size
 Pool and Conv. More...
 
int stride
 Pool, Conv, and Upsample. More...
 
int pad
 Pool and Conv. More...
 
double eta
 Gradient descent rate. More...
 
double eta_min
 Current gradient descent rate. More...
 
double momentum
 Momentum for gradient descent. More...
 
double decay
 Weight decay for gradient descent. More...
 
double probability
 Usage depends on layer implementation. More...
 
double scale
 Usage depends on layer implementation. More...
 
_Bool evolve_weights
 Ability to evolve weights. More...
 
_Bool evolve_neurons
 Ability to evolve number of units. More...
 
_Bool evolve_functions
 Ability to evolve activation function. More...
 
_Bool evolve_eta
 Ability to evolve gradient descent rate. More...
 
_Bool evolve_connect
 Ability to evolve weight connectivity. More...
 
_Bool sgd_weights
 Ability to update weights with gradient descent. More...
 
struct ArgsLayer * next
 Next layer parameters. More...
 

Detailed Description

Parameters for initialising a neural network layer.

Definition at line 31 of file neural_layer_args.h.

Field Documentation

◆ channels

int ArgsLayer::channels

Pool, Conv, and Upsample.

Definition at line 41 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_inputs(), layer_args_json_import_inputs(), layer_args_load(), layer_args_save(), layer_args_validate_inputs(), neural_create(), neural_layer_avgpool_init(), neural_layer_convolutional_init(), neural_layer_dropout_init(), neural_layer_maxpool_init(), neural_layer_noise_init(), and neural_layer_upsample_init().

◆ decay

double ArgsLayer::decay

Weight decay for gradient descent.

Definition at line 48 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_sgd(), layer_args_json_import_sgd(), layer_args_load(), layer_args_save(), neural_layer_connected_init(), neural_layer_convolutional_init(), neural_layer_lstm_init(), pred_neural_ae_to_classifier(), and pred_neural_expand().

◆ eta

double ArgsLayer::eta

Gradient descent rate.

Definition at line 45 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_sgd(), layer_args_json_import_sgd(), layer_args_load(), layer_args_save(), neural_layer_connected_init(), neural_layer_convolutional_init(), neural_layer_lstm_init(), pred_neural_ae_to_classifier(), pred_neural_expand(), and pred_neural_param_defaults().

◆ eta_min

double ArgsLayer::eta_min

Current gradient descent rate.

Definition at line 46 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_sgd(), layer_args_json_import_sgd(), layer_args_load(), layer_args_save(), neural_layer_connected_init(), neural_layer_convolutional_init(), pred_neural_ae_to_classifier(), and pred_neural_expand().

◆ evolve_connect

_Bool ArgsLayer::evolve_connect

Ability to evolve weight connectivity.

Definition at line 55 of file neural_layer_args.h.

Referenced by act_neural_param_defaults(), cond_neural_param_defaults(), layer_args_copy(), layer_args_init(), layer_args_json_export_evo(), layer_args_json_import_evo(), layer_args_load(), layer_args_opt(), layer_args_save(), pred_neural_ae_to_classifier(), pred_neural_expand(), and pred_neural_param_defaults().

◆ evolve_eta

_Bool ArgsLayer::evolve_eta

Ability to evolve gradient descent rate.

Definition at line 54 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_sgd(), layer_args_json_import_sgd(), layer_args_load(), layer_args_opt(), layer_args_save(), pred_neural_ae_to_classifier(), pred_neural_expand(), and pred_neural_param_defaults().

◆ evolve_functions

_Bool ArgsLayer::evolve_functions

Ability to evolve activation function.

Definition at line 53 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_evo(), layer_args_json_import_evo(), layer_args_load(), layer_args_opt(), layer_args_save(), and pred_neural_expand().

◆ evolve_neurons

_Bool ArgsLayer::evolve_neurons

Ability to evolve number of units.

Definition at line 52 of file neural_layer_args.h.

Referenced by act_neural_param_defaults(), cond_neural_param_defaults(), layer_args_copy(), layer_args_init(), layer_args_json_export_evo(), layer_args_json_import_evo(), layer_args_load(), layer_args_opt(), layer_args_save(), layer_args_validate(), and pred_neural_param_defaults().

◆ evolve_weights

_Bool ArgsLayer::evolve_weights

Ability to evolve weights.

Definition at line 51 of file neural_layer_args.h.

Referenced by act_neural_param_defaults(), cond_neural_param_defaults(), layer_args_copy(), layer_args_init(), layer_args_json_export_evo(), layer_args_json_import_evo(), layer_args_load(), layer_args_opt(), layer_args_save(), pred_neural_ae_to_classifier(), pred_neural_expand(), and pred_neural_param_defaults().

◆ function

int ArgsLayer::function

Activation function.

Definition at line 37 of file neural_layer_args.h.

Referenced by act_neural_param_defaults(), cond_neural_param_defaults(), layer_args_copy(), layer_args_init(), layer_args_json_export_activation(), layer_args_json_import_activation(), layer_args_load(), layer_args_save(), neural_layer_connected_init(), neural_layer_convolutional_init(), neural_layer_lstm_init(), neural_layer_recurrent_init(), pred_neural_expand(), and pred_neural_param_defaults().

◆ height

int ArgsLayer::height

Pool, Conv, and Upsample.

Definition at line 39 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_inputs(), layer_args_json_import_inputs(), layer_args_load(), layer_args_save(), layer_args_validate_inputs(), neural_create(), neural_layer_avgpool_init(), neural_layer_convolutional_init(), neural_layer_dropout_init(), neural_layer_maxpool_init(), neural_layer_noise_init(), and neural_layer_upsample_init().

◆ max_neuron_grow

int ArgsLayer::max_neuron_grow

Maximum number neurons to add per mutation event.

Definition at line 36 of file neural_layer_args.h.

Referenced by act_neural_param_defaults(), cond_neural_param_defaults(), layer_args_copy(), layer_args_init(), layer_args_json_export_evo(), layer_args_json_import_evo(), layer_args_load(), layer_args_save(), layer_args_validate(), neural_layer_connected_init(), neural_layer_convolutional_init(), neural_layer_lstm_init(), and pred_neural_param_defaults().

◆ momentum

double ArgsLayer::momentum

Momentum for gradient descent.

Definition at line 47 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_sgd(), layer_args_json_import_sgd(), layer_args_load(), layer_args_save(), neural_layer_connected_init(), neural_layer_convolutional_init(), neural_layer_lstm_init(), pred_neural_ae_to_classifier(), pred_neural_expand(), and pred_neural_param_defaults().

◆ n_init

int ArgsLayer::n_init

Initial number of units / neurons / filters.

Definition at line 34 of file neural_layer_args.h.

Referenced by act_neural_param_defaults(), cond_neural_param_defaults(), layer_args_copy(), layer_args_init(), layer_args_json_import(), layer_args_load(), layer_args_save(), layer_args_validate(), neural_create(), neural_layer_connected_init(), neural_layer_convolutional_init(), neural_layer_lstm_init(), neural_layer_recurrent_init(), and pred_neural_param_defaults().

◆ n_inputs

int ArgsLayer::n_inputs

Number of inputs.

Definition at line 33 of file neural_layer_args.h.

Referenced by xcsf.utils.viz.DGPViz::__init__(), act_neural_param_defaults(), act_neural_param_json_import(), cond_neural_param_defaults(), cond_neural_param_json_import(), xcsf.utils.viz.DGPViz::draw(), layer_args_copy(), layer_args_init(), layer_args_json_export_inputs(), layer_args_json_import_inputs(), layer_args_load(), layer_args_save(), layer_args_validate_inputs(), neural_create(), neural_layer_connected_init(), neural_layer_dropout_init(), neural_layer_lstm_init(), neural_layer_noise_init(), neural_layer_recurrent_init(), neural_layer_softmax_init(), pred_neural_expand(), pred_neural_param_defaults(), and pred_neural_param_json_import().

◆ n_max

int ArgsLayer::n_max

Maximum number of units / neurons.

Definition at line 35 of file neural_layer_args.h.

Referenced by act_neural_param_defaults(), cond_neural_param_defaults(), layer_args_copy(), layer_args_init(), layer_args_json_export_evo(), layer_args_json_import_evo(), layer_args_load(), layer_args_save(), layer_args_validate(), neural_layer_connected_init(), neural_layer_convolutional_init(), neural_layer_lstm_init(), neural_layer_recurrent_init(), and pred_neural_param_defaults().

◆ next

struct ArgsLayer* ArgsLayer::next

Next layer parameters.

Definition at line 57 of file neural_layer_args.h.

Referenced by act_neural_param_defaults(), act_neural_param_json_import(), cond_neural_param_defaults(), cond_neural_param_json_import(), layer_args_free(), layer_args_init(), layer_args_json_export(), layer_args_length(), layer_args_load(), layer_args_save(), layer_args_tail(), layer_args_validate(), neural_create(), pred_neural_param_defaults(), and pred_neural_param_json_import().

◆ pad

int ArgsLayer::pad

Pool and Conv.

Definition at line 44 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_inputs(), layer_args_json_import_inputs(), layer_args_load(), layer_args_save(), neural_layer_convolutional_init(), and neural_layer_maxpool_init().

◆ probability

double ArgsLayer::probability

Usage depends on layer implementation.

Definition at line 49 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_scale(), layer_args_json_import_scale(), layer_args_load(), layer_args_save(), neural_layer_dropout_init(), and neural_layer_noise_init().

◆ recurrent_function

int ArgsLayer::recurrent_function

Recurrent activation function.

Definition at line 38 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_activation(), layer_args_json_import_activation(), layer_args_load(), layer_args_save(), and neural_layer_lstm_init().

◆ scale

double ArgsLayer::scale

Usage depends on layer implementation.

Definition at line 50 of file neural_layer_args.h.

Referenced by act_neural_param_defaults(), layer_args_copy(), layer_args_init(), layer_args_json_export_scale(), layer_args_json_import_scale(), layer_args_load(), layer_args_save(), neural_layer_noise_init(), and neural_layer_softmax_init().

◆ sgd_weights

_Bool ArgsLayer::sgd_weights

Ability to update weights with gradient descent.

Definition at line 56 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_sgd(), layer_args_json_import_sgd(), layer_args_load(), layer_args_opt(), layer_args_save(), pred_neural_ae_to_classifier(), pred_neural_expand(), and pred_neural_param_defaults().

◆ size

int ArgsLayer::size

Pool and Conv.

Definition at line 42 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_inputs(), layer_args_json_import_inputs(), layer_args_load(), layer_args_save(), neural_layer_convolutional_init(), and neural_layer_maxpool_init().

◆ stride

int ArgsLayer::stride

Pool, Conv, and Upsample.

Definition at line 43 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_inputs(), layer_args_json_import_inputs(), layer_args_load(), layer_args_save(), neural_layer_convolutional_init(), neural_layer_maxpool_init(), and neural_layer_upsample_init().

◆ type

int ArgsLayer::type

Layer type: CONNECTED, DROPOUT, etc.

Definition at line 32 of file neural_layer_args.h.

Referenced by act_neural_param_defaults(), cond_neural_param_defaults(), layer_args_copy(), layer_args_init(), layer_args_json_export_activation(), layer_args_json_export_inputs(), layer_args_json_export_scale(), layer_args_json_import(), layer_args_load(), layer_args_save(), layer_args_validate_inputs(), layer_init(), neural_create(), neural_layer_lstm_init(), neural_layer_recurrent_init(), and pred_neural_param_defaults().

◆ width

int ArgsLayer::width

Pool, Conv, and Upsample.

Definition at line 40 of file neural_layer_args.h.

Referenced by layer_args_copy(), layer_args_init(), layer_args_json_export_inputs(), layer_args_json_import_inputs(), layer_args_load(), layer_args_save(), layer_args_validate_inputs(), neural_create(), neural_layer_avgpool_init(), neural_layer_convolutional_init(), neural_layer_dropout_init(), neural_layer_maxpool_init(), neural_layer_noise_init(), and neural_layer_upsample_init().

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