XCSF data structure. More...

#include <xcsf.h>

Collaboration diagram for XCSF:

Data Fields
struct Set	pset
	Population set. More...

struct Set	prev_pset
	Previously stored population set. More...

struct Set	mset
	Match set. More...

struct Set	aset
	Action set. More...

struct Set	kset
	Kill set. More...

struct Set	prev_aset
	Previous action set. More...

struct ArgsAct *	act
	Action parameters. More...

struct ArgsCond *	cond
	Condition parameters. More...

struct ArgsPred *	pred
	Prediction parameters. More...

struct ArgsEA *	ea
	EA parameters. More...

struct EnvVtbl const *	env_vptr
	Functions acting on environments. More...

void *	env
	Environment structure (for built-in problems) More...

double	error
	Average system error. More...

double	mset_size
	Average match set size. More...

double	aset_size
	Average action set size. More...

double	mfrac
	Generalisation measure. More...

double	prev_reward
	Reward from previous step in a multi-step trial. More...

double	prev_pred
	Payoff prediction made on the previous step. More...

double *	pa
	Prediction array (stores fitness weighted predictions) More...

double *	nr
	Prediction array (stores total fitness) More...

double *	prev_state
	Environment state on the previous step. More...

double *	cover
	Values to return for a prediction instead of covering. More...

int	time
	Current number of EA executions. More...

int	pa_size
	Prediction array size. More...

int	x_dim
	Number of problem input variables. More...

int	y_dim
	Number of problem output variables. More...

int	n_actions
	Number of class labels / actions. More...

bool	explore
	Whether the system is currently exploring or exploiting. More...

double(*	loss_ptr )(const struct XCSF , const double , const double *)
	Error function. More...

double	GAMMA
	Discount factor for multi-step reward. More...

double	P_EXPLORE
	Probability of exploring vs. exploiting. More...

double	ALPHA
	Linear coefficient used to calculate classifier accuracy. More...

double	BETA
	Learning rate for updating error, fitness, and set size. More...

double	DELTA
	Fraction of population to increase deletion vote. More...

double	E0
	Target error under which classifier accuracy is set to 1. More...

double	INIT_ERROR
	Initial classifier error value. More...

double	INIT_FITNESS
	Initial classifier fitness value. More...

double	NU
	Exponent used in calculating classifier accuracy. More...

double	HUBER_DELTA
	Delta parameter for Huber loss calculation. More...

int	OMP_NUM_THREADS
	Number of threads for parallel processing. More...

int	MAX_TRIALS
	Number of problem instances to run in one experiment. More...

int	PERF_TRIALS
	Number of problem instances to avg performance output. More...

int	POP_SIZE
	Maximum number of micro-classifiers in the population. More...

int	LOSS_FUNC
	Which loss/error function to apply. More...

int	TELETRANSPORTATION
	Maximum steps for a multi-step problem. More...

int	THETA_DEL
	Min experience before fitness used during deletion. More...

int	M_PROBATION
	Trials since creation a cl must match at least 1 input. More...

int	THETA_SUB
	Minimum experience of a classifier to become a subsumer. More...

int	RANDOM_STATE
	Random number seed. More...

bool	POP_INIT
	Pop initially empty or filled with random conditions. More...

bool	SET_SUBSUMPTION
	Whether to perform match set subsumption. More...

bool	STATEFUL
	Whether classifiers should retain state across trials. More...

bool	COMPACTION
	if sys err < E0: largest of 2 roulette spins deleted More...

char *	population_file
	Name of a JSON file containing an initial pop. More...

Detailed Description

XCSF data structure.

Definition at line 85 of file xcsf.h.

Field Documentation

◆ act

struct ArgsAct* XCSF::act

Action parameters.

Definition at line 92 of file xcsf.h.

◆ ALPHA

double XCSF::ALPHA

Linear coefficient used to calculate classifier accuracy.

Definition at line 118 of file xcsf.h.

◆ aset

struct Set XCSF::aset

Action set.

Definition at line 79 of file xcsf.h.

◆ aset_size

double XCSF::aset_size

Average action set size.

Definition at line 100 of file xcsf.h.

Referenced by XCS::get_aset_size().

◆ BETA

double XCSF::BETA

Learning rate for updating error, fitness, and set size.

Definition at line 119 of file xcsf.h.

◆ COMPACTION

bool XCSF::COMPACTION

if sys err < E0: largest of 2 roulette spins deleted

Definition at line 139 of file xcsf.h.

◆ cond

struct ArgsCond* XCSF::cond

Condition parameters.

Definition at line 93 of file xcsf.h.

◆ cover

double* XCSF::cover

Values to return for a prediction instead of covering.

Definition at line 107 of file xcsf.h.

Referenced by XCS::fit(), XCS::load_validation_data(), XCS::predict(), XCS::score(), and XCS::set_cover().

◆ DELTA

double XCSF::DELTA

Fraction of population to increase deletion vote.

Definition at line 120 of file xcsf.h.

◆ E0

double XCSF::E0

Target error under which classifier accuracy is set to 1.

Definition at line 121 of file xcsf.h.

◆ ea

struct ArgsEA* XCSF::ea

EA parameters.

Definition at line 95 of file xcsf.h.

◆ env

void* XCSF::env

Environment structure (for built-in problems)

Definition at line 97 of file xcsf.h.

◆ env_vptr

struct EnvVtbl const* XCSF::env_vptr

Functions acting on environments.

Definition at line 96 of file xcsf.h.

◆ error

double XCSF::error

Average system error.

Definition at line 98 of file xcsf.h.

Referenced by XCS::error().

◆ explore

bool XCSF::explore

Whether the system is currently exploring or exploiting.

Definition at line 113 of file xcsf.h.

◆ GAMMA

double XCSF::GAMMA

Discount factor for multi-step reward.

Definition at line 116 of file xcsf.h.

◆ HUBER_DELTA

double XCSF::HUBER_DELTA

Delta parameter for Huber loss calculation.

Definition at line 125 of file xcsf.h.

◆ INIT_ERROR

double XCSF::INIT_ERROR

Initial classifier error value.

Definition at line 122 of file xcsf.h.

◆ INIT_FITNESS

double XCSF::INIT_FITNESS

Initial classifier fitness value.

Definition at line 123 of file xcsf.h.

◆ kset

struct Set XCSF::kset

Kill set.

Definition at line 79 of file xcsf.h.

◆ LOSS_FUNC

int XCSF::LOSS_FUNC

Which loss/error function to apply.

Definition at line 130 of file xcsf.h.

◆ loss_ptr

double(* XCSF::loss_ptr) (const struct XCSF *, const double *, const double *)

Error function.

Definition at line 114 of file xcsf.h.

◆ M_PROBATION

int XCSF::M_PROBATION

Trials since creation a cl must match at least 1 input.

Definition at line 133 of file xcsf.h.

◆ MAX_TRIALS

int XCSF::MAX_TRIALS

Number of problem instances to run in one experiment.

Definition at line 127 of file xcsf.h.

Referenced by XCS::fit().

◆ mfrac

double XCSF::mfrac

Generalisation measure.

Definition at line 101 of file xcsf.h.

Referenced by XCS::get_mfrac(), and XCS::update_metrics().

◆ mset

struct Set XCSF::mset

Match set.

Definition at line 79 of file xcsf.h.

◆ mset_size

double XCSF::mset_size

Average match set size.

Definition at line 99 of file xcsf.h.

Referenced by XCS::get_mset_size(), and XCS::update_metrics().

◆ n_actions

int XCSF::n_actions

Number of class labels / actions.

Definition at line 112 of file xcsf.h.

Referenced by XCS::fit().

◆ nr

double* XCSF::nr

Prediction array (stores total fitness)

Definition at line 105 of file xcsf.h.

◆ NU

double XCSF::NU

Exponent used in calculating classifier accuracy.

Definition at line 124 of file xcsf.h.

◆ OMP_NUM_THREADS

int XCSF::OMP_NUM_THREADS

Number of threads for parallel processing.

Definition at line 126 of file xcsf.h.

◆ P_EXPLORE

double XCSF::P_EXPLORE

Probability of exploring vs. exploiting.

Definition at line 117 of file xcsf.h.

◆ pa

double* XCSF::pa

Prediction array (stores fitness weighted predictions)

Definition at line 104 of file xcsf.h.

◆ pa_size

int XCSF::pa_size

Prediction array size.

Definition at line 109 of file xcsf.h.

Referenced by XCS::load_validation_data(), XCS::predict(), and XCS::set_cover().

◆ PERF_TRIALS

int XCSF::PERF_TRIALS

Number of problem instances to avg performance output.

Definition at line 128 of file xcsf.h.

Referenced by XCS::fit().

◆ POP_INIT

bool XCSF::POP_INIT

Pop initially empty or filled with random conditions.

Definition at line 136 of file xcsf.h.

◆ POP_SIZE

int XCSF::POP_SIZE

Maximum number of micro-classifiers in the population.

Definition at line 129 of file xcsf.h.

◆ population_file

char* XCSF::population_file

Name of a JSON file containing an initial pop.

Definition at line 140 of file xcsf.h.

◆ pred

struct ArgsPred* XCSF::pred

Prediction parameters.

Definition at line 94 of file xcsf.h.

◆ prev_aset

struct Set XCSF::prev_aset

Previous action set.

Definition at line 79 of file xcsf.h.

◆ prev_pred

double XCSF::prev_pred

Payoff prediction made on the previous step.

Definition at line 103 of file xcsf.h.

◆ prev_pset

struct Set XCSF::prev_pset

Previously stored population set.

Definition at line 79 of file xcsf.h.

◆ prev_reward

double XCSF::prev_reward

Reward from previous step in a multi-step trial.

Definition at line 102 of file xcsf.h.

◆ prev_state

double* XCSF::prev_state

Environment state on the previous step.

Definition at line 106 of file xcsf.h.

◆ pset

struct Set XCSF::pset

Population set.

Definition at line 79 of file xcsf.h.

Referenced by XCS::get_pset_mean_cond_connections(), XCS::get_pset_mean_cond_layers(), XCS::get_pset_mean_cond_neurons(), XCS::get_pset_mean_cond_size(), XCS::get_pset_mean_pred_connections(), XCS::get_pset_mean_pred_eta(), XCS::get_pset_mean_pred_layers(), XCS::get_pset_mean_pred_neurons(), XCS::get_pset_mean_pred_size(), XCS::get_pset_num(), XCS::get_pset_size(), XCS::json_export(), and XCS::update_metrics().

◆ RANDOM_STATE

int XCSF::RANDOM_STATE

Random number seed.

Definition at line 135 of file xcsf.h.

◆ SET_SUBSUMPTION

bool XCSF::SET_SUBSUMPTION

Whether to perform match set subsumption.

Definition at line 137 of file xcsf.h.

◆ STATEFUL

bool XCSF::STATEFUL

Whether classifiers should retain state across trials.

Definition at line 138 of file xcsf.h.

◆ TELETRANSPORTATION

int XCSF::TELETRANSPORTATION

Maximum steps for a multi-step problem.

Definition at line 131 of file xcsf.h.

◆ THETA_DEL

int XCSF::THETA_DEL

Min experience before fitness used during deletion.

Definition at line 132 of file xcsf.h.

◆ THETA_SUB

int XCSF::THETA_SUB

Minimum experience of a classifier to become a subsumer.

Definition at line 134 of file xcsf.h.

◆ time

int XCSF::time

Current number of EA executions.

Definition at line 108 of file xcsf.h.

Referenced by XCS::get_time().

◆ x_dim

int XCSF::x_dim

Number of problem input variables.

Definition at line 110 of file xcsf.h.

Referenced by XCS::decision(), XCS::fit(), XCS::load_input(), and XCS::predict().

◆ y_dim

int XCSF::y_dim

Number of problem output variables.

Definition at line 111 of file xcsf.h.

Referenced by XCS::get_cover(), and XCS::load_input().

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

xcsf.h

Data Fields

Detailed Description

Field Documentation

◆ act

◆ ALPHA

◆ aset

◆ aset_size

◆ BETA

◆ COMPACTION

◆ cond

◆ cover

◆ DELTA

◆ E0

◆ ea

◆ env

◆ env_vptr

◆ error

◆ explore

◆ GAMMA

◆ HUBER_DELTA

◆ INIT_ERROR

◆ INIT_FITNESS

◆ kset

◆ LOSS_FUNC

◆ loss_ptr

◆ M_PROBATION

◆ MAX_TRIALS

◆ mfrac

◆ mset

◆ mset_size

◆ n_actions

◆ nr

◆ NU

◆ OMP_NUM_THREADS

◆ P_EXPLORE

◆ pa

◆ pa_size

◆ PERF_TRIALS

◆ POP_INIT

◆ POP_SIZE

◆ population_file

◆ pred

◆ prev_aset

◆ prev_pred

◆ prev_pset

◆ prev_reward

◆ prev_state

◆ pset

◆ RANDOM_STATE

◆ SET_SUBSUMPTION

◆ STATEFUL

◆ TELETRANSPORTATION

◆ THETA_DEL

◆ THETA_SUB

◆ time

◆ x_dim

◆ y_dim