% Perceptron Example 1: 2-input neuron, 4 input vectors
% author: G. Dempsey, Bradley University
% January 2003

% learning rate = 0.5
eta = 0.5

% threshold = 0
%
% Patterns:

P1 = [0.3 0.7]';     % A1
P2 = [0.7 0.3]';     % A2
P3 = [-0.6 0.3]';    % B1
P4 = [-0.2 -0.8]';   % B2

% Define Pattern matrix

d1=['display input vectors'];
disp(d1)

P= [P1 P2 P3 P4]

%
% Target Outputs
%

T1 = 1;               % class 1
T2 = 1;               % class 1
T3 = -1;              % class 2
T4 = -1;              % class 2

% define Target matrix

d1=['display target matrix'];
disp(d1)

T= [T1 T2 T3 T4]

d2=['press any key to continue'];
disp(d2)
pause


%
% Initialize synaptic weights
%
d1=['display initial weight matrix'];
disp(d1)

W = [-0.6 0.8]'

d1=['press any key to start training'];
disp(d1)
pause


% Set number of pattern presentations to 4

n = 4;

epoch = 0;              % initialize epochs to zero
                        % define as each training input

% OK Train the neuron


for i = 1:n

   for j = 1:4
    
   v(j) = W'*P(:,j);   % calculate neuron input for pattern j
   y(j) = sign(v(j));  % calculate neuron output for pattern j
   e(j) = T(j)-y(j);   % calculate output error for pattern j

   W = W + eta*e(j)*P(:,j); % adapt weights

   epoch=epoch+1;

   end

d1=['display input, output, error, weights after each epoch'];
disp(d1)

v       % input
y       % output
e       % error
W       % weights
epoch   % epochs

d3=['press any key to continue training'];
disp(d3)
pause

end