% This program uses the weights obtained from the training process with
%TestMultiClass.m to classify slightly corrupted images.

clear all
close all

%TestMultiClass; % call TestMultiClass.m to retrieve weights W1 and W2.
TestMultiClass_sol; % call TestMultiClass.m to retrieve weights W1 and W2.
clc

% The input set consists of five 5x5 pixel squares.
%0: white pixel; 1: black pixel.
X = zeros(5, 5, 5);
X(:, :, 1) = [0 1 1 1 1;
              0 0 0 0 1;
              0 0 0 1 1;
              0 0 1 0 1;
              0 0 1 0 0]; %corrupted 7.
X(:, :, 2) = [0 0 0 1 0;
              0 0 1 0 0;
              0 1 1 0 0;
              1 0 0 1 0;
              0 0 0 1 0]; %corrupted 4.
X(:, :, 3) = [0 1 1 1 0;
              0 0 0 1 0;
              0 1 1 1 0;
              1 1 0 0 0;
              0 1 1 1 0]; %corrupted 2.
X(:, :, 4) = [1 1 1 1 0;
              1 0 0 0 0;
              1 0 0 0 1;
              1 0 0 0 1;
              0 1 1 1 1]; %corrupted 6.
X(:, :, 5) = [0 1 1 1 1;
              0 1 0 0 0;
              0 1 1 1 0;
              0 0 0 1 0;
              1 1 1 1 0]; %corrupted 5.

% Inference:
classes = 7;
N = size(X,3);
y = zeros(N,classes); %5 because of the one-hot encoding method.
for k = 1:N
    x = reshape(X(:, :, k), 25, 1);
    v1 = W1*x;
    y1 = Sigmoid(v1);
    v = W2*y1;
    y(k,:) = Softmax(v);
end

disp('Results:');
disp('   [network_output]:');
disp(y)

%% Showing images:
for i = 1:N
    compareImages(X(:,:,i), y(i,:));
end