individual programs in the population More...

#include <individual.h>

Collaboration diagram for FT::Pop::Individual:

[legend]

Public Types
typedef Array< bool, Dynamic, Dynamic, RowMajor >	ArrayXXb

Public Member Functions
	Individual ()

void	initialize (const Parameters &params, bool random, int id=0)
	copy assignment More...

MatrixXf	out (const Data &d, bool predict=false)
	calculate program output matrix Phi More...

MatrixXf	out_trace (const Data &d, vector< Trace > &stack_trace)
	calculate program output while maintaining stack trace More...

MatrixXf	state_to_phi (State &state)
	converts program states to output matrices More...

shared_ptr< CLabels >	fit (const Data &d, const Parameters &params, bool &pass)
	fits an ML model to the data after transformation More...

shared_ptr< CLabels >	fit (const Data &d, const Parameters &params)
	fits an ML model to the data after transformation More...

shared_ptr< CLabels >	fit_tune (const Data &d, const Parameters &params, bool set_default=false)
	fits and tunes an ML model to the data after transformation More...

void	tune (const Data &d, const Parameters &params)
	tunes an ML model to the data after transformation More...

shared_ptr< CLabels >	predict (const Data &d)

VectorXf	predict_vector (const Data &d)

ArrayXXf	predict_proba (const Data &d)

string	get_eqn ()
	return symbolic representation of program More...

vector< string >	get_features ()
	return vectorized representation of program More...

string	program_str () const
	return program name list More...

void	set_rank (unsigned r)
	setting and getting from individuals vector More...

int	size () const
	return size of program More...

int	get_n_params ()
	get number of params in program More...

unsigned int	get_dim ()
	grab sub-tree locations given starting point. More...

int	check_dominance (const Individual &b) const
	check whether this dominates b. More...

void	set_obj (const vector< string > &)
	set obj vector given a string of objective names More...

unsigned int	set_complexity ()
	calculate program complexity and return it. More...

unsigned int	get_complexity () const
	get the program complexity without updating it. More...

void	clone (Individual &cpy, bool sameid=true) const
	clone this individual More...

Individual	clone ()

void	set_id (unsigned i)

void	set_parents (const vector< Individual > &parents)
	set parent ids using parents More...

void	set_parents (const vector< int > &parents)
	set parent ids using id values More...

vector< float >	get_p () const
	get probabilities of variation More...

float	get_p (const size_t i, bool normalize=true) const
	get inverted weight probability for pogram location i More...

vector< float >	get_p (const vector< size_t > &locs, bool normalize=false) const
	get probability of variation for program locations locs More...

void	set_p (const vector< float > &weights, const float &fb, const bool softmax_norm=false)
	set probabilities More...

std::map< char, size_t >	get_max_state_size ()
	get maximum stack size needed for evaluation. More...

void	save (string filename)
	save individual as a json object. More...

void	load (string filename)
	load individual from a file. More...

Public Attributes
NodeVector	program
	executable data structure More...

MatrixXf	Phi
	transformation output of program More...

VectorXf	yhat
	current output More...

VectorXf	error
	training error More...

shared_ptr< ML >	ml
	ML model, trained on Phi. More...

float	fitness
	aggregate fitness score More...

float	fitness_v
	aggregate validation fitness score More...

float	fairness
	aggregate fairness score More...

float	fairness_v
	aggregate validation fairness score More...

vector< float >	w
	weights from ML training on program output More...

vector< float >	p
	probability of variation of subprograms More...

unsigned int	dim
	dimensionality of individual More...

vector< float >	obj
	objectives for use with Pareto selection More...

unsigned int	dcounter
	number of individuals this dominates More...

vector< unsigned int >	dominated
	individual indices this dominates More...

unsigned int	rank
	pareto front rank More...

float	crowd_dist
	crowding distance on the Pareto front More...

unsigned int	complexity
	the complexity of the program. More...

vector< char >	dtypes
	the data types of each column of the More...

unsigned	id
	tracking id More...

vector< int >	parent_id
	ids of parents More...

string	eqn
	equation form of the program More...

Detailed Description

individual programs in the population

Definition at line 31 of file individual.h.

Member Typedef Documentation

◆ ArrayXXb

typedef Array<bool, Dynamic, Dynamic, RowMajor> FT::Pop::Individual::ArrayXXb

Definition at line 173 of file individual.h.

Constructor & Destructor Documentation

◆ Individual()

FT::Pop::Individual::Individual ( )

Definition at line 11 of file individual.cc.

Member Function Documentation

◆ check_dominance()

int FT::Pop::Individual::check_dominance ( const Individual & b ) const

check whether this dominates b.

Definition at line 898 of file individual.cc.

◆ clone() [1/2]

Individual FT::Pop::Individual::clone ( )

Definition at line 71 of file individual.cc.

◆ clone() [2/2]

void FT::Pop::Individual::clone	(	Individual &	cpy,
		bool	sameid = `true`
	)		const

clone this individual

Definition at line 82 of file individual.cc.

◆ fit() [1/2]

shared_ptr< CLabels > FT::Pop::Individual::fit	(	const Data &	d,
		const Parameters &	params
	)

fits an ML model to the data after transformation

Definition at line 264 of file individual.cc.

◆ fit() [2/2]

shared_ptr< CLabels > FT::Pop::Individual::fit	(	const Data &	d,
		const Parameters &	params,
		bool &	pass
	)

fits an ML model to the data after transformation

Definition at line 234 of file individual.cc.

◆ fit_tune()

shared_ptr< CLabels > FT::Pop::Individual::fit_tune	(	const Data &	d,
		const Parameters &	params,
		bool	set_default = `false`
	)

fits and tunes an ML model to the data after transformation

Definition at line 1019 of file individual.cc.

◆ get_complexity()

unsigned int FT::Pop::Individual::get_complexity ( ) const

get the program complexity without updating it.

Definition at line 109 of file individual.cc.

◆ get_dim()

unsigned int FT::Pop::Individual::get_dim ( )

grab sub-tree locations given starting point.

get program dimensionality

Output:

@return the dimensionality, i.e. number of outputs, 
of a program. the dimensionality is equal to the number of 
times the program arities are fully
satisfied.

Definition at line 873 of file individual.cc.

◆ get_eqn()

string FT::Pop::Individual::get_eqn ( )

return symbolic representation of program

Definition at line 748 of file individual.cc.

◆ get_features()

vector< string > FT::Pop::Individual::get_features ( )

return vectorized representation of program

Definition at line 817 of file individual.cc.

◆ get_max_state_size()

std::map< char, size_t > FT::Pop::Individual::get_max_state_size ( )

get maximum stack size needed for evaluation.

Definition at line 995 of file individual.cc.

◆ get_n_params()

int FT::Pop::Individual::get_n_params ( )

get number of params in program

Definition at line 96 of file individual.cc.

◆ get_p() [1/3]

vector< float > FT::Pop::Individual::get_p ( ) const

get probabilities of variation

Definition at line 122 of file individual.cc.

◆ get_p() [2/3]

float FT::Pop::Individual::get_p	(	const size_t	i,
		bool	normalize = `true`
	)		const

get inverted weight probability for pogram location i

Parameters

i	index in program
normalize	(true): normalizes the probability by the size of the subprogram. Useful when the total probability over the program nodes should sum to 1.

Returns: weight associated with node

Definition at line 175 of file individual.cc.

◆ get_p() [3/3]

vector< float > FT::Pop::Individual::get_p	(	const vector< size_t > &	locs,
		bool	normalize = `false`
	)		const

get probability of variation for program locations locs

Parameters

locs	program indices to return probabilities for.
normalize	(false): normalize probabilities by size of subprogram

Returns: float vector of probabilities

Definition at line 217 of file individual.cc.

◆ initialize()

void FT::Pop::Individual::initialize	(	const Parameters &	params,
		bool	random,
		int	id = `0`
	)

copy assignment

Definition at line 27 of file individual.cc.

◆ load()

void FT::Pop::Individual::load ( string filename )

load individual from a file.

Definition at line 1061 of file individual.cc.

◆ out()

MatrixXf FT::Pop::Individual::out	(	const Data &	d,
		bool	predict = `false`
	)

calculate program output matrix Phi

Parameters

d	Data structure
params	Feat parameters
predict	if true, this guarantees nodes like split do not get trained

Returns: Phi: n_features x n_samples transformation

Definition at line 391 of file individual.cc.

◆ out_trace()

MatrixXf FT::Pop::Individual::out_trace	(	const Data &	d,
		vector< Trace > &	stack_trace
	)

calculate program output while maintaining stack trace

Parameters

X	n_features x n_samples data
Z	longitudinal nodes for samples
y	target data
	Feat parameters

Returns: Phi: n_features x n_samples transformation

Definition at line 544 of file individual.cc.

◆ predict()

shared_ptr< CLabels > FT::Pop::Individual::predict ( const Data & d )

generates prediction on data using transformation and ML predict.

Parameters

drop_idx if specified, the phi output at drop_idx is set to zero, effectively removing its output from the transformation. used in semantic crossover.

Definition at line 271 of file individual.cc.

◆ predict_proba()

ArrayXXf FT::Pop::Individual::predict_proba ( const Data & d )

Definition at line 293 of file individual.cc.

◆ predict_vector()

VectorXf FT::Pop::Individual::predict_vector ( const Data & d )

Definition at line 311 of file individual.cc.

◆ program_str()

string FT::Pop::Individual::program_str ( ) const

return program name list

Definition at line 982 of file individual.cc.

◆ save()

void FT::Pop::Individual::save ( string filename )

save individual as a json object.

Definition at line 1050 of file individual.cc.

◆ set_complexity()

unsigned int FT::Pop::Individual::set_complexity ( )

calculate program complexity and return it.

Definition at line 967 of file individual.cc.

◆ set_id()

void FT::Pop::Individual::set_id ( unsigned i )

Definition at line 112 of file individual.cc.

◆ set_obj()

void FT::Pop::Individual::set_obj ( const vector< string > & objectives )

set obj vector given a string of objective names

Input: objectives: vector of strings naming objectives.

Definition at line 936 of file individual.cc.

◆ set_p()

void FT::Pop::Individual::set_p	(	const vector< float > &	weights,
		const float &	fb,
		const bool	softmax_norm = `false`
	)

set probabilities

Definition at line 124 of file individual.cc.

◆ set_parents() [1/2]

void FT::Pop::Individual::set_parents ( const vector< Individual > & parents )

set parent ids using parents

Definition at line 114 of file individual.cc.

◆ set_parents() [2/2]

void FT::Pop::Individual::set_parents ( const vector< int > & parents )

inline

set parent ids using id values

Definition at line 150 of file individual.h.

◆ set_rank()

void FT::Pop::Individual::set_rank ( unsigned r )

setting and getting from individuals vector

set rank

Definition at line 90 of file individual.cc.

◆ size()

int FT::Pop::Individual::size ( ) const

return size of program

Definition at line 93 of file individual.cc.

◆ state_to_phi()

MatrixXf FT::Pop::Individual::state_to_phi ( State & state )

converts program states to output matrices

Definition at line 316 of file individual.cc.

◆ tune()

void FT::Pop::Individual::tune	(	const Data &	d,
		const Parameters &	params
	)

tunes an ML model to the data after transformation

Member Data Documentation

◆ complexity

unsigned int FT::Pop::Individual::complexity

the complexity of the program.

Definition at line 50 of file individual.h.

◆ crowd_dist

float FT::Pop::Individual::crowd_dist

crowding distance on the Pareto front

Definition at line 49 of file individual.h.

◆ dcounter

unsigned int FT::Pop::Individual::dcounter

number of individuals this dominates

Definition at line 46 of file individual.h.

◆ dim

unsigned int FT::Pop::Individual::dim

dimensionality of individual

Definition at line 44 of file individual.h.

◆ dominated

vector<unsigned int> FT::Pop::Individual::dominated

individual indices this dominates

Definition at line 47 of file individual.h.

◆ dtypes

vector<char> FT::Pop::Individual::dtypes

the data types of each column of the

Definition at line 51 of file individual.h.

◆ eqn

string FT::Pop::Individual::eqn

equation form of the program

Definition at line 55 of file individual.h.

◆ error

VectorXf FT::Pop::Individual::error

training error

Definition at line 36 of file individual.h.

◆ fairness

float FT::Pop::Individual::fairness

aggregate fairness score

Definition at line 40 of file individual.h.

◆ fairness_v

float FT::Pop::Individual::fairness_v

aggregate validation fairness score

Definition at line 41 of file individual.h.

◆ fitness

float FT::Pop::Individual::fitness

aggregate fitness score

Definition at line 38 of file individual.h.

◆ fitness_v

float FT::Pop::Individual::fitness_v

aggregate validation fitness score

Definition at line 39 of file individual.h.

◆ id

unsigned FT::Pop::Individual::id

tracking id

Definition at line 53 of file individual.h.

◆ ml

shared_ptr<ML> FT::Pop::Individual::ml

ML model, trained on Phi.

Definition at line 37 of file individual.h.

◆ obj

vector<float> FT::Pop::Individual::obj

objectives for use with Pareto selection

Definition at line 45 of file individual.h.

◆ p

vector<float> FT::Pop::Individual::p

probability of variation of subprograms

Definition at line 43 of file individual.h.

◆ parent_id

vector<int> FT::Pop::Individual::parent_id

ids of parents

Definition at line 54 of file individual.h.

◆ Phi

MatrixXf FT::Pop::Individual::Phi

transformation output of program

Definition at line 34 of file individual.h.

◆ program

NodeVector FT::Pop::Individual::program

executable data structure

Definition at line 33 of file individual.h.

◆ rank

unsigned int FT::Pop::Individual::rank

pareto front rank

Definition at line 48 of file individual.h.

◆ w

vector<float> FT::Pop::Individual::w

weights from ML training on program output

Definition at line 42 of file individual.h.

◆ yhat

VectorXf FT::Pop::Individual::yhat

current output

Definition at line 35 of file individual.h.

The documentation for this class was generated from the following files:

Public Types

Public Member Functions

Public Attributes

Detailed Description

Member Typedef Documentation

◆ ArrayXXb

Constructor & Destructor Documentation

◆ Individual()

Member Function Documentation

◆ check_dominance()

◆ clone() [1/2]

◆ clone() [2/2]

◆ fit() [1/2]

◆ fit() [2/2]

◆ fit_tune()

◆ get_complexity()

◆ get_dim()

◆ get_eqn()

◆ get_features()

◆ get_max_state_size()

◆ get_n_params()

◆ get_p() [1/3]

◆ get_p() [2/3]

◆ get_p() [3/3]

◆ initialize()

◆ load()

◆ out()

◆ out_trace()

◆ predict()

◆ predict_proba()

◆ predict_vector()

◆ program_str()

◆ save()

◆ set_complexity()

◆ set_id()

◆ set_obj()

◆ set_p()

◆ set_parents() [1/2]

◆ set_parents() [2/2]

◆ set_rank()

◆ size()

◆ state_to_phi()

◆ tune()

Member Data Documentation

◆ complexity

◆ crowd_dist

◆ dcounter

◆ dim

◆ dominated

◆ dtypes

◆ eqn

◆ error

◆ fairness

◆ fairness_v

◆ fitness

◆ fitness_v

◆ id

◆ ml

◆ obj

◆ p

◆ parent_id

◆ Phi

◆ program

◆ rank

◆ w

◆ yhat