Adaboost 的自定义学习器函数

Posted 2023-03-12

【中文标题】Adaboost 的自定义学习器函数

【英文标题】：Custom learner function for Adaboost

【发布时间】：2018-02-06 17:41:40

【问题描述】：

我正在使用 Adaboost 来拟合分类问题。我们可以做到以下几点：

ens = fitensemble(X, Y, 'AdaBoostM1', 100, 'Tree')

现在“树”是学习者，我们可以将其更改为“判别”或“KNN”。每个学习者使用特定的Template Object Creation Function。更多信息here。

是否可以创建自己的函数并将其用作学习器？怎么做？

【问题讨论】：

不容易。例如，考虑您的树学习器，您需要一个 templateTree() 和一个 ClassificationTree 类，您都可以在 %matlabroot%/toolbox/stats/clas-s-reg 中找到它们。您可以对clas-s-reg.learning.FitTemplate.make() 所需的通用 API 进行逆向工程，但我认为这是一个为期一周的项目。说了这么多，你有想要实现的特定学习器吗？

它确实是一个自定义函数，而不是文学的通用学习器。

【参考方案1】：

我打开 templateTree.m 和 templateKNN.m 看看 MATLAB 是如何定义模板对象创建函数的。

function temp = templateKNN(varargin)
     clas-s-reg.learning.FitTemplate.catchType(varargin:);
     temp = clas-s-reg.learning.FitTemplate.make('KNN','type','classification',varargin:);
end

和

function temp = templateTree(varargin)
   temp = clas-s-reg.learning.FitTemplate.make('Tree',varargin:);
end

说明 MATLAB 在 FitTemplate 中有一个名为 make 的函数，如果你打开这个 m 文件，你会看到：

   function temp = make(method,varargin)
        % Check the type of the required argument
        if ~ischar(method)
            error(message('stats:clas-s-reg:learning:FitTemplate:make:BadArgs'));
        end

        % Extract type (classification or regression)
        args = 'type';
        defs =     '';
        [usertype,~,modelArgs] = ...
            internal.stats.parseArgs(args,defs,varargin:);

        % Check usertype
        if ~isempty(usertype)
            usertype = gettype(usertype);
        end

        % Method
        namesclass = clas-s-reg.learning.classificationModels();
        namesreg = clas-s-reg.learning.regressionModels();
        [tfclass,locclass] = ismember(lower(method),lower(namesclass));
        [tfreg,locreg] = ismember(lower(method),lower(namesreg));
        if ~tfclass && ~tfreg
            error(message('stats:clas-s-reg:learning:FitTemplate:make:UnknownMethod', method));
        end
        if     tfclass && tfreg
            method = namesclasslocclass; % can get it from namesreg too
            type = usertype;

            % If type is not passed for an ensemble method, try to
            % figure it out from learner types. This is useful for
            % users who want to type
            %   fitensemble(X,Y,'Subspace',100,'Discriminant')
            % instead of
            %   fitensemble(X,Y,'Subspace',100,'Discriminant','type','classification')
            if isempty(type) && ismember(method,clas-s-reg.learning.ensembleModels())
                [learners,~,~] = internal.stats.parseArgs('learners',,modelArgs:);
                if ischar(learners) || isa(learners,'clas-s-reg.learning.FitTemplate')
                    learners = learners;
                elseif ~iscell(learners)
                    error(message('stats:clas-s-reg:learning:FitTemplate:make:BadLearnerTemplates'));
                end
                L = numel(learners);
                % The user can pass several learner templates, and some
                % of these learners may be appropriate for
                % classification, some for regression, and some for
                % both. The ensemble type cannot be determined
                % unambiguously unless if all learners are appropriate
                % for one type of learning *only*. For example, in 12a
                %   t1 = ClassificationDiscriminant.template
                %   t2 = ClassificationKNN.template
                %   fitensemble(X,Y,'Subspace',10,t1 t2)
                % is going to work because both discriminant and k-NN
                % can be used for classification only. If you want to
                % mix discriminant and tree, you have to specify the
                % ensemble type explicitly:
                %   t1 = ClassificationDiscriminant.template
                %   t2 = ClassificationTree.template
                %   fitensemble(X,Y,'Bag',10,t1 t2,'type','classification')
                types = zeros(L,1); % -1 for regression and 1 for classification
                for l=1:L
                    meth = learnersl;
                    if isa(meth,'clas-s-reg.learning.FitTemplate')
                        meth = meth.Method;
                    end
                    isc = ismember(lower(meth),lower(namesclass));
                    isr = ismember(lower(meth),lower(namesreg));
                    if ~isc && ~isr
                        error(message('stats:clas-s-reg:learning:FitTemplate:make:UnknownMethod', meth));
                    end
                    types(l) = isc - isr;
                end
                if     all(types==1)
                    type = 'classification';
                elseif all(types==-1)
                    type = 'regression';
                end
            end
        elseif tfclass
            method = namesclasslocclass;
            type = 'classification';
        else
            method = namesreglocreg;
            type = 'regression';
        end

        % Make sure the type is consistent
        if ~isempty(usertype) && ~strcmp(usertype,type)
            error(message('stats:clas-s-reg:learning:FitTemplate:make:UserTypeMismatch', method, usertype));
        end

        % Make template
        temp = clas-s-reg.learning.FitTemplate(method,modelArgs);
        temp = fillIfNeeded(temp,type);
    end

您必须更改此功能。