Add files via upload

BEEFCN.m

@@ -0,0 +1,138 @@
+function bestfis=BEEFCN(fis,data)
+% Variables
+p0=GettingFuzzyParameters(fis);
+Problem.CostFunction=@(x) FuzzyCost(x,fis,data);
+Problem.nVar=numel(p0);
+alpha=1;
+Problem.VarMin=-(10^alpha);
+Problem.VarMax=10^alpha;
+% Bees Algorithm Parameters
+Params.MaxIt=15;
+Params.nScoutBee = 10;                                  % Number of Scout Bees
+Params.nSelectedSite = round(0.5*Params.nScoutBee);     % Number of Selected Sites
+Params.nEliteSite = round(0.4*Params.nSelectedSite);    % Number of Selected Elite Sites
+Params.nSelectedSiteBee = round(0.5*Params.nScoutBee);  % Number of Recruited Bees for Selected Sites
+Params.nEliteSiteBee = 2*Params.nSelectedSiteBee;       % Number of Recruited Bees for Elite Sites
+Params.r = 0.1*(Problem.VarMax-Problem.VarMin);         % Neighborhood Radius
+Params.rdamp = 0.95;                                    % Neighborhood Radius Damp Rate
+% Starting Bees Algorithm
+results=Runbee(Problem,Params);
+% Getting the Results
+p=results.BestSol.Position.*p0;
+bestfis=FuzzyParameters(fis,p);
+end
+%% Bees
+function results=Runbee(Problem,Params)
+disp('Starting Bees Algorithm Training :)');
+% Cost Function
+CostFunction=Problem.CostFunction;
+% Number of Decision Variables
+nVar=Problem.nVar;
+% Size of Decision Variables Matrixv
+VarSize=[1 nVar];
+% Lower Bound of Variables
+VarMin=Problem.VarMin;
+% Upper Bound of Variables
+VarMax=Problem.VarMax;
+% Some Change
+if isscalar(VarMin) && isscalar(VarMax)
+dmax = (VarMax-VarMin)*sqrt(nVar);
+else
+dmax = norm(VarMax-VarMin);
+end
+%% Bees Algorithm Parameters
+MaxIt=Params.MaxIt;
+nScoutBee = Params.nScoutBee;                % Number of Scout Bees
+nSelectedSite = Params.nSelectedSite;        % Number of Selected Sites
+nEliteSite = Params.nEliteSite;              % Number of Selected Elite Sites
+nSelectedSiteBee = Params.nSelectedSiteBee;  % Number of Recruited Bees for Selected Sites
+nEliteSiteBee = Params.nEliteSiteBee;        % Number of Recruited Bees for Elite Sites
+r = Params.r;                                % Neighborhood Radius
+rdamp = Params.rdamp;                        % Neighborhood Radius Damp Rate
+%% Second Stage
+% Empty Bee Structure
+empty_bee.Position = [];
+empty_bee.Cost = [];
+% Initialize Bees Array
+bee = repmat(empty_bee, nScoutBee, 1);
+% Create New Solutions
+for i = 1:nScoutBee
+bee(i).Position = unifrnd(VarMin, VarMax, VarSize);
+bee(i).Cost = CostFunction(bee(i).Position);
+end
+% Sort
+[~, SortOrder] = sort([bee.Cost]);
+bee = bee(SortOrder);
+% Update Best Solution Ever Found
+BestSol = bee(1);
+% Array to Hold Best Cost Values
+BestCost = zeros(MaxIt, 1);
+%% Bees Algorithm Main Body
+for it = 1:MaxIt
+% Elite Sites
+for i = 1:nEliteSite
+bestnewbee.Cost = inf;
+for j = 1:nEliteSiteBee
+newbee.Position = PerformBeeDance(bee(i).Position, r);
+newbee.Cost = CostFunction(newbee.Position);
+if newbee.Cost<bestnewbee.Cost
+bestnewbee = newbee;
+end
+end
+if bestnewbee.Cost<bee(i).Cost
+bee(i) = bestnewbee;
+end
+end
+% Selected Non-Elite Sites
+for i = nEliteSite+1:nSelectedSite
+bestnewbee.Cost = inf;
+for j = 1:nSelectedSiteBee
+newbee.Position = PerformBeeDance(bee(i).Position, r);
+newbee.Cost = CostFunction(newbee.Position);
+if newbee.Cost<bestnewbee.Cost
+bestnewbee = newbee;
+end
+end
+if bestnewbee.Cost<bee(i).Cost
+bee(i) = bestnewbee;
+end
+end
+% Non-Selected Sites
+for i = nSelectedSite+1:nScoutBee
+bee(i).Position = unifrnd(VarMin, VarMax, VarSize);
+bee(i).Cost = CostFunction(bee(i).Position);
+end
+% Sort
+[~, SortOrder] = sort([bee.Cost]);
+bee = bee(SortOrder);
+% Update Best Solution Ever Found
+BestSol = bee(1);
+% Store Best Cost Ever Found
+BestCost(it) = BestSol.Cost;
+% Display Iteration Information
+disp(['In Iteration Number ' num2str(it) '  Bees Algorithm Fittest Value Is = ' num2str(BestCost(it))]);
+% Damp Neighborhood Radius
+r = r*rdamp;
+end
+disp('Bees Algorithm Came To End :)');
+% Store Res
+results.BestSol=BestSol;
+results.BestCost=BestCost;
+% Plot Bees Algorithm Training Stages
+set(gcf, 'Position',  [600, 300, 500, 400])
+plot(BestCost,':',...
+'LineWidth',2,...
+'MarkerSize',8,...
+'MarkerEdgeColor','g',...
+'Color',[0.1,0.9,0.1]);
+title('Bees Algorithm Training')
+xlabel('Bees Algorithm Iteration Number','FontSize',10,...
+'FontWeight','bold','Color','m');
+ylabel('Bees Algorithm Fittest Value','FontSize',10,...
+'FontWeight','bold','Color','m');
+legend({'Bees Algorithm Train'});
+end
+
+
+
+

Bees CNN Algorithm.m

@@ -0,0 +1,178 @@
+%% Bees CNN Algorithm (A Fuzzy Evolutionary Deep Leaning) - Created in 20 Jan 2022 by Seyed Muhammad Hossein Mousavi
+% It is possible to fit deep learning weights and bias using evolutionary
+% algorithm, right after training stage. Here, CNN is used to classify 8
+% face classes. After CNN train, initial fuzzy model is created to aid the
+% learning process. Finally, CNN network weights (from Fully Connected Layer)
+% trains using Bees algorithm
+% to be fitted in a nature inspired manner (here behavior of Bees). You can
+% used your data with any number of samples and classes. Remember, code's
+% parameters are adjusted for this data and if you want to replace your
+% data you may have to change the parameters. Image data is in 64*64 size and
+% in 2 dimensions and stored in 'CNNDat' folder. So, important parameters 
+% are as below:
+% 1.
+% 'numTrainFiles' = you have to change this based on number of your samples
+% in each class. for example if each class has 120 sample, 90 is good
+% enough as 90 samples considered for train and others for test.
+% 2.
+% 'imageInputLayer' = it is size of your image data like [64 64 1]
+% 3.
+% 'fullyConnectedLayer' = it is number of your classes like (8)
+% 4.
+% 'MaxEpochs' = the more the better and more computation run time like 40
+% 5.
+% 'ClusNum' = Fuzzy C Means (FCM) Cluster Number like 3 or 4 is nice
+% 6.
+% These two are from "BEEFCN.m" function :
+% 'Params.MaxIt' = it is iteration number in Bees algorithm. 20 is good
+% 'Params.nScoutBee' = it is population number in Bees algorithm. Like 10.
+% ------------------------------------------------ 
+% Feel free to contact me if you find any problem using the code: 
+% Author: SeyedMuhammadHosseinMousavi
+% My Email: [email protected] 
+% My Google Scholar: https://scholar.google.com/citations?user=PtvQvAQAAAAJ&hl=en 
+% My GitHub: https://github.com/SeyedMuhammadHosseinMousavi?tab=repositories 
+% My ORCID: https://orcid.org/0000-0001-6906-2152 
+% My Scopus: https://www.scopus.com/authid/detail.uri?authorId=57193122985 
+% My MathWorks: https://www.mathworks.com/matlabcentral/profile/authors/9763916#
+% my RG: https://www.researchgate.net/profile/Seyed-Mousavi-17
+% ------------------------------------------------ 
+% Hope it help you, enjoy the code and wish me luck :)
+
+%% Cleaning
+ clear;
+ clc;
+ warning('off');
+
+%% CNN Deep Neural Network
+% Load the deep sample data as an image datastore. 
+deepDatasetPath = fullfile('CNNDat');
+imds = imageDatastore(deepDatasetPath, ...
+    'IncludeSubfolders',true, ...
+    'LabelSource','foldernames');
+% Divide the data into training and validation data sets
+numTrainFiles = 90;
+[imdsTrain,imdsValidation] = splitEachLabel(imds,numTrainFiles,'randomize');
+% Define the convolutional neural network architecture.
+layers = [
+% Image Input Layer An imageInputLayer 
+    imageInputLayer([64 64 1])
+% Convolutional Layer 
+convolution2dLayer(3,8,'Padding','same')
+% Batch Normalization 
+    batchNormalizationLayer
+% ReLU Layer The batch
+    reluLayer
+% Max Pooling Layer  
+    % More values means less weights
+    maxPooling2dLayer(4,'Stride',4)
+    %------------------------------
+    convolution2dLayer(3,8,'Padding','same')
+    batchNormalizationLayer
+    reluLayer
+    maxPooling2dLayer(5,'Stride',5)
+    convolution2dLayer(3,8,'Padding','same')
+    batchNormalizationLayer
+    reluLayer
+% Fully Connected Layer (Number of Classes) 
+    fullyConnectedLayer(8)
+% Softmax Layer 
+    softmaxLayer
+% Classification Layer The final layer 
+    classificationLayer];
+% Specify the training options
+options = trainingOptions('sgdm', ...
+    'InitialLearnRate',0.001, ...
+    'MaxEpochs',20, ...
+    'Shuffle','every-epoch', ...
+    'ValidationData',imdsValidation, ...
+    'ValidationFrequency',8, ...
+    'Verbose',false, ...
+    'Plots','training-progress');
+% Train the network 
+[net,info]= trainNetwork(imdsTrain,layers,options);
+
+%% Bees Algorithm Weight Fitting
+% Converting Serial Network to an Object
+netobj = net.saveobj;
+% Extracting Fully Connected Layer's Weights To Evolve
+FullConn=netobj.Layers(13, 1).Weights;
+netbias=netobj.Layers(13, 1).Bias;
+
+%% Data for Each Weight
+sizefinal=size(FullConn);
+sizefinal=sizefinal(1,1);
+for i=1:sizefinal
+Inputs=FullConn(i,:);
+Targets=Inputs;
+data.Inputs=Inputs;
+data.Targets=Targets;
+datam{i}=JustLoad(data);
+end;
+
+%% Making Basic Fuzzy Model for Each Class Weight
+% Fuzzy C Means (FCM) Cluster Number
+ClusNum=3; 
+% Creating Initial Fuzzy Model to Employ for Each Class Weight
+for i=1:sizefinal
+fism{i}=GenerateFuzzy(datam{i},ClusNum);
+end
+
+%% Tarining Bees Algorithm
+% Fitting Fully Connected Layer's Weights with Bees Algorithm
+for i=1:sizefinal
+disp(['Bees Are Working on Weights of Class # (' num2str(i) ')']);
+BeesFISm{i}=BEEFCN(fism{i},datam{i}); 
+end;
+
+%% Train Output Extraction
+for i=1:sizefinal
+TrTar{i}=datam{i}.TrainTargets;
+TrInp{i}=datam{i}.TrainInputs;
+TrainOutputs{i}=evalfis(TrInp{i},BeesFISm{i});
+end;
+% Train Errors
+for i=1:sizefinal
+tmp=datam{i};
+tt=tmp.TrainTargets;
+tp=TrainOutputs{i};
+Errors{i}=tt-tp;
+MSE{i}=mean(Errors{i}.^2);
+RMSE{i}=sqrt(MSE{i});  
+error_mean{i}=mean(Errors{i});
+error_std{i}=std(Errors{i});
+end;
+% Convereting Output Cell to Matrix
+for i=1:sizefinal
+EvolvedFullConn(i,:)=TrainOutputs{i}';
+end;
+
+%% Replacing Evolved Weights
+netobj.Layers(13, 1).Weights=EvolvedFullConn;
+% New Network
+Newnet=netobj.Layers;
+% Converting Network to Serial Network
+BeesNet = assembleNetwork(Newnet);
+
+%% Predict The Labels 
+% Common CNN Accuracy
+YPred = classify(net,imdsValidation);
+YValidation = imdsValidation.Labels;
+CNNaccuracy = sum(YPred == YValidation)/numel(YValidation);
+% Bees CNN Accuracy
+YPredbee = classify(BeesNet,imdsValidation);
+YValidationbee = imdsValidation.Labels;
+Beesaccuracy = sum(YPredbee == YValidationbee)/numel(YValidationbee);
+
+%% Confusion Matrix
+figure;
+plotconfusion(YPred,YValidation);
+title(['CNN Accuracy  =  ' num2str(CNNaccuracy)]);
+figure;
+plotconfusion(YPredbee,YValidationbee);
+title(['Bees-CNN Accuracy  =  ' num2str(Beesaccuracy)]);
+
+%% Statistics
+fprintf('The CNN Accuracy Is =  %0.4f.\n',CNNaccuracy*100)
+fprintf('The Bees CNN Accuracy Is =  %0.4f.\n',Beesaccuracy*100)
+

FuzzyCost.m

@@ -0,0 +1,21 @@
+function [z, out]=FuzzyCost(x,fis,data)
+    MinAbs=1e-5;
+    if any(abs(x)<MinAbs)
+        S=(abs(x)<MinAbs);
+        x(S)=MinAbs.*sign(x(S));
+    end
+    p0=GettingFuzzyParameters(fis);
+    p=x.*p0;  
+    fis=FuzzyParameters(fis,p);   
+    x=data.TrainInputs;
+    t=data.TrainTargets;
+    y=evalfis(x,fis);    
+    e=t-y;    
+    MSE=mean(e(:).^2);
+    RMSE=sqrt(MSE);    
+    z=RMSE;    
+    out.fis=fis;
+    out.MSE=MSE;
+    out.RMSE=RMSE;
+
+end

FuzzyParameters.m

@@ -0,0 +1,20 @@
+function fis=FuzzyParameters(fis,p)
+    nInput=numel(fis.input);
+    for i=1:nInput
+        nMF=numel(fis.input(i).mf);
+        for j=1:nMF
+            k=numel(fis.input(i).mf(j).params);
+            fis.input(i).mf(j).params=p(1:k);
+            p(1:k)=[];
+        end
+    end
+    nOutput=numel(fis.output);
+    for i=1:nOutput
+        nMF=numel(fis.output(i).mf);
+        for j=1:nMF
+            k=numel(fis.output(i).mf(j).params);
+            fis.output(i).mf(j).params=p(1:k);
+            p(1:k)=[];
+        end
+    end
+end

GenerateFuzzy.m

@@ -0,0 +1,16 @@
+
+function fis=GenerateFuzzy(data,nCluster)
+    if ~exist('nCluster','var')
+        nCluster='auto';
+    end
+    x=data.TrainInputs;
+    t=data.TrainTargets;
+    % Important Params
+    fcm_U=2;
+    fcm_MaxIter=100;
+    fcm_MinImp=1e-5;
+    %
+    fcm_Display=false;
+    fcm_options=[fcm_U fcm_MaxIter fcm_MinImp fcm_Display];
+    fis=genfis3(x,t,'sugeno',nCluster,fcm_options);
+end

GettingFuzzyParameters.m

@@ -0,0 +1,17 @@
+function p=GettingFuzzyParameters(fis)
+    p=[];
+    nInput=numel(fis.input);
+    for i=1:nInput
+        nMF=numel(fis.input(i).mf);
+        for j=1:nMF
+            p=[p fis.input(i).mf(j).params];
+        end
+    end
+    nOutput=numel(fis.output);
+    for i=1:nOutput
+        nMF=numel(fis.output(i).mf);
+        for j=1:nMF
+            p=[p fis.output(i).mf(j).params];
+        end
+    end  
+end