AdaBoost

An implementation of the AdaBoost MH (Adaptive Boosting) algorithm for classification.

:- use_module('path/to/.../src/methods/adaboost/adaboost.pl').

%% usage example
TrainData = [5.1,3.5,1.4, 4.9,3.0,1.4, 4.7,3.2,1.3, 4.6,3.1,1.5],
TestData = [3,2,0, 5,1,4, 0,0,4, 3,3,5, 0,5,5, 2,5,5],
adaboost_initModelWithTraining(TrainData, 3, [0,1,0,1], 2, perceptron, 50, 0.0001),
adaboost_classify(TestData, 3, PredictionList, ProbabilitiesList, _).

Available Predicates

links/resources

adaboost_initModelWithTraining/7

Needs to be called first before all other predicates exept initModelNoTraining!

Initiates the Adaboostmodel and trains it, so classify can be used immediately.

%% predicate definition 
adaboost_initModelWithTraining(MatList, MatRows, VecList, NumClasses, Learner, Iterations, Tolerance) :-
        NumClasses >= 0,
        Iterations >= 0,
        Tolerance > 0,
        convert_list_to_float_array(MatList, MatRows, array(Xsize, Xrownum, X)),
        convert_list_to_float_array(VecList, array(Ysize, Y)),
        initModelWithTrainingI(X, Xsize, Xrownum, Y, Ysize, NumClasses, Learner, Iterations, Tolerance).

%% foreign c++ predicate definition 
foreign(initModelWithTraining, c, initModelWithTrainingI(       +pointer(float_array), +integer, +integer, 
                                                                +pointer(float_array), +integer, 
                                                                +integer, 
                                                                +string, 
                                                                +integer , +float32)).

Parameters

Name	Type	Description	Default
data	+matrix	Dataset for training AdaBoost	-
labels	+vector	Labels for the training set	-
numClasses	+integer	The number of classes to classify data into	-
learner	+string	The type of weak learner to use: “decision_stump”, or “perceptron”	“decision_stump”
iterations	+integer	The maximum number of boosting iterations to be run (0 will run until convergence.)	100
tolerance	+float	The tolerance for change in values of the weighted error during training	1e-6

adaboost_initModelNoTraining/2

Needs to be called first before all other predicates exept initModelWithTraining!

Initiates the Adaboostmodel but doesn’t train it, so train has to be used first before classify can be used.

%% predicate definition
adaboost_initModelNoTraining(Tolerance, Learner) :-
        Tolerance > 0,
        initModelNoTrainingI(Tolerance, Learner).

%% foreign c++ predicate definition 
foreign(initModelNoTraining, c, initModelNoTrainingI(+float32, +string)).

Parameters

Name	Type	Description	Default
tolerance	+float	The tolerance for change in values of the weighted error during training	1e-6
learner	+string	The type of weak learner to use: “decision_stump”, or “perceptron”	“decision_stump”

adaboost_classify/5

Classifies the given data into the number of classes the model was trained for.

%% predicate definition 
adaboost_classify(TestList, TestRows, PredicList, ProbsList, ZRows) :-
        convert_list_to_float_array(TestList, TestRows, array(Xsize, Xrownum, X)),
        classifyI(X, Xsize, Xrownum, Y, Ysize, Z, ZCols, ZRows),
        convert_float_array_to_list(Y, Ysize, PredicList),
        convert_float_array_to_2d_list(Z, ZCols, ZRows, ProbsList).

%% foreign c++ predicate definition 
foreign(classify, c, classifyI( +pointer(float_array), +integer, +integer, 
                                -pointer(float_array), -integer, 
                                -pointer(float_array), -integer, -integer)).

Parameters

Name	Type	Description	Default
test	+matrix	Test dataset	-
predictions	-vector	Predicted labels for the test set	-
probabilities	-matrix	Predicted class probabilities for each point in the test set	-

adaboost_numClasses/1

Returns the amount of classes defined in the model for classification.

%% predicate definition 
adaboost_numClasses(ClassesNum) :-
        numClassesI(ClassesNum).

%% foreign c++ predicate definition 
foreign(numClasses, c, numClassesI([-integer])).

Parameters

Name	Type	Description	Default
classesNum	-integer	Number of classes	-

adaboost_getTolerance/1

Returns the tolerance of the model.

%% predicate definition
adaboost_getTolerance(Tolerance) :-
        getToleranceI(Tolerance).

%% foreign c++ predicate definition 
foreign(getTolerance, c, getToleranceI([-float32])).

Parameters

Name	Type	Description	Default
tolerance	-float32	returns the current tolerance	-

adaboost_modifyTolerance/1

Modifies the tolerance of the model.

%% predicate definition 
adaboost_modifyTolerance(NewTolerance) :-
        NewTolerance > 0,
        modifyToleranceI(NewTolerance).

%% foreign c++ predicate definition 
foreign(modifyTolerance, c, modifyToleranceI(+float32)).

Parameters

Name	Type	Description	Default
tolerance	+float32	changes the tolerance	1e-6

adaboost_train/8

Kurze beschreibung mit evt. hinweisen

%% predicate definition 
adaboost_train(MatList, MatRows, VecList, NumClasses, Learner, Iterations, Tolerance, Error) :-
        NumClasses >= 0,
        Iterations >= 0,
        Tolerance >= 0,
        convert_list_to_float_array(MatList, MatRows, array(Xsize, Xrownum, X)),
        convert_list_to_float_array(VecList, array(Ysize, Y)),
        trainI(X, Xsize, Xrownum, Y, Ysize, NumClasses, Learner, Iterations, Tolerance, Error).

%% foreign c++ predicate definition 
foreign(train, c, trainI(   +pointer(float_array), +integer, +integer, 
                            +pointer(float_array), +integer, 
                            +integer, 
                            +string, 
                            +integer , +float32, 
                            [-float32])).

Parameters

Name	Type	Description	Default
data	+matrix	Dataset for training AdaBoost	-
labels	+vector	Labels for the training set	-
numClasses	+integer	The number of classes to classify data into	-
learner	+string	The type of weak learner to use: “decision_stump”, or “perceptron”	“decision_stump”
iterations	+integer	The maximum number of boosting iterations to be run (0 will run until convergence.)	100
tolerance	+float	The tolerance for change in values of the weighted error during training	1e-6
error	-float	double upper bound training error	-

Connected Links/Resources

If you want a more detailed explanation, then go to the python documentation. There is most of the time a good explanation on how the methods work and about what the parameters do.

added some of the links from the python documentation

Comments

Please register or sign in to add a comment.