JAYA Optimization based Feed-Forward Network Matlab Code

JAYA Optimization based Feed-Forward Neural Network's Matlab Code, created by Sana Mujeeb

JAYA Optimization based Feed-Forward Neural Network's Matlab Code, created by Sana Mujeeb. 14th August, 2022.

Website: https://sites.google.com/view/sanamujeeb/home?authuser=0

Google Scholar: https://scholar.google.com/citations?user=oOSExv4AAAAJ&hl=en

Researchgate: https://www.researchgate.net/profile/Sana-Mujeeb

Proposed by: Wang S, Rao RV, Chen P, Zhang Y, Liu A, Wei L. Abnormal breast detection in mammogram images by feed-forward neural network trained by Jaya algorithm. Fundamenta Informaticae. 2017 Jan 1;151(1-4):191-211.

Any errors and suggestions of improvements in his code are welcome. If you have any questions regarding this code, then please ask me.

Below is the complete code other than the attached file. Just copy it and paste on a m file and run the code, it has one file.

Enjoy!!!

% *************************************************************************************************************

% Source Code of JAYA Optimization based Feed-Forward

% Neural Network By Sana Mujeeb, 14h August, 2022

% Cite: Wang S, Rao RV, Chen P, Zhang Y, Liu A, Wei L. Abnormal breast detection in

% mammogram images by feed-forward neural network trained by Jaya algorithm.

% Fundamenta Informaticae. 2017 Jan 1;151(1-4):191-211.

% *************************************************************************************************************

% Enjoy JAYA-ANN!

clc;

% Generating random correlated data

mu = 50;

sigma = 5;

M = mu + sigma * randn(300, 2);

R = [1, 0.75; 0.75, 1];

L = chol(R);

M = M*L;

x = M(:,1); % Example Inputs, Replace by your data inputs for your own experiments

y = M(:,2); % Example labels, Replace by your data labels for your own experiments

%% JAYA algorithms

%% Problem Definition

pop = 30; % Population size

% Min-max normalization of data

m = max(x); mn = min(x); mm = m-mn;

X = ((x-mn)/mm); Y = ((x-mn)/mm);

% 90%:10% splitting of data for training and testing

sz = (ceil(size(X,1))*0.9);

inputs = (X(1:sz))';

targets = (Y(1:sz))';

XTest = (X(sz+1:end))';

YTest = Y(sz+1:end)';

% number of neurons

n = 4;

tic;

% create a neural network

net = feedforwardnet(n);

% configure the neural network for this dataset

net = configure(net, inputs, targets);

% Denormalizaion and Prediction by FNN

FNN_Pred = ((net(XTest))' * mm) + mn;

sz = n^2 + n + n + 1; % Number of design variables i.e., no. of weights in FNN

maxGen = 30; % Maximum number of iterations

mini = repmat(-1,1,sz); % Lower Bound of Variables

maxi = ones(1,sz); % Upper Bound of Variables

objective = @(x) NMSE(x, net, inputs, targets); % Cost Function

%% initialize

[row,var] = size(mini);

x = zeros(pop,var);

fnew = zeros(pop,1);

f = zeros(pop,1);

fopt= zeros(pop,1);

xopt=zeros(1,var);

%% Generation and Initialize the positions

for i=1:var

x(:,i) = mini(i)+(maxi(i)-mini(i))*rand(pop,1);

end

for i=1:pop

f(i) = objective(x(i,:));

end

%% Main Loop

gen=1;

fprintf('Best Cost per Iteration of JAYA Opimization Algorithm \n');

while(gen <= maxGen)

[row,col]=size(x);

[t,tindex]=min(f);

Best=x(tindex,:);

[w,windex]=max(f);

worst=x(windex,:);

xnew=zeros(row,col);

for i=1:row

for j=1:col

xnew(i,j)=(x(i,j))+rand*(Best(j)-abs(x(i,j))) - (worst(j)-abs(x(i,j))); %

end

end

for i=1:row

xnew(i,:) = max(min(xnew(i,:),maxi),mini);

fnew(i,:) = objective(xnew(i,:));

end

for i=1:pop

if(fnew(i)<f(i))

x(i,:) = xnew(i,:);

f(i) = fnew(i);

end

end

fnew = []; xnew = [];

[fopt(gen),ind] = min(f);

xopt(gen,:)= x(ind,:);

gen = gen+1;

disp(['Iteration No. = ',num2str(gen-1), ', Best Cost = ',num2str(min(f))])

end

%%

[val,ind] = min(fopt);

Fes = pop*ind;

disp(['Optimum value = ',num2str(val,10)])

figure;

plot(fopt,'LineWidth', 2);

xlabel('Itteration');

ylabel('Best Cost');

legend('JAYA');

disp(' ' );

% Setting optimized weights and bias in network

net = setwb(net, Best');

% Denormalizaion and Prediction by JAYA_FNN

JAYA_FNN_Pred = ((net(XTest))' * mm) + mn;

YTest = (YTest * mm) + mn;

JAYA_FNN_Execution_Time_Seconds = toc

% Plotting prediction results

figure;

plot(YTest,'LineWidth',2, 'Marker','diamond', 'MarkerSize',8);

hold on;

plot(FNN_Pred, 'LineWidth',2, 'Marker','x', 'MarkerSize',8);

plot(JAYA_FNN_Pred, 'LineWidth',2, 'Marker','pentagram', 'MarkerSize',8);

title('JAYA Optimization based Feed-Forward Neural Network');

xlabel('Time Interval');

ylabel('Values');

legend('Actual Values', 'FNN Predictions', 'JAYA-FNN Predictions');

hold off;

% Performance Evaluaion of FNN and JAYA-FNN

fprintf('Performance Evaluaion of FNN and JAYA-FNN using Normalized Root Mean Square Error \n');

NRMSE_FNN = (abs( sqrt( mean(mean((FNN_Pred - YTest).^2) )) )) / (max(YTest)-min(YTest))

NRMSE_JAYA_FNN = (abs( sqrt( mean(mean((JAYA_FNN_Pred - YTest).^2) ) ) )) / (max(YTest)-min(YTest))

% Objective Function for minimizing normalized mean square error of FNN by

% updation of nework's weights and biases

function [f] = NMSE(wb, net, input, target)

% wb is the weights and biases row vector obtained from the genetic algorithm.

% It must be transposed when transferring the weights and biases to the network net.

net = setwb(net, wb');

% The net output matrix is given by net(input). The corresponding error matrix is given by

error = target - net(input);

% The mean squared error normalized by the mean target variance is

f = (mean(error.^2)/mean(var(target',1)));

% It is independent of the scale of the target components and related to the Rsquare statistic via

% Rsquare = 1 - NMSEcalc ( see Wikipedia)

end