mexicanhat.m

內涵模糊理論與類神經網路的程式碼...提供初學者做研究參考

% ==========================================================
% 
%           Neural Networks A Classroom Approach
%                     Satish Kumar
%             Copyright Tata McGraw Hill, 2004
%
%       MATLAB code that implements the Mexican hat 
%       network connectivity
%             Reference: Table 12.6;Page 551
%
% ==========================================================

% Mexican Hat Network Simulation

leradius = 5;	% Excitation radius
liwidth = 10;	% Inhibition radius
interactlen = leradius + liwidth +1;
max = 10;		% Maximum signal value
excit = 0.1;		% a
inhibit = -0.05;	% b
feedback = 1.5;	% gamma

for j=1:50			% Generate the Mexican hat connectivity
   for i=1:50		% 50 x 50 weights
      indexdif = abs(j-i);
      if (indexdif < interactlen)
         if (indexdif < leradius+1)
            w(j,i) = excit;
         else
            w(j,i) = inhibit;
         end
      else
         w(j,i) = 0;
      end
   end
end

index=1:1:50;
input = sin(pi.*index/50);	% Set up input vector
s=zeros(50);					% Initialize signals

figure;

subplot(1,2,1);
hold on

for t=1:15
   for i =1:50	% Compute activations
      activation(i) = input(i);
      for j=1:50
         activation(i) = activation(i) + feedback*w(j,i)*s(j);
      end
   end
   
   for i=1:50	% compute signals
      if (activation(i) > max) s(i) = max;
      elseif (activation(i) < 0) s(i) = 0;
      else s(i) = activation(i);
      end
   end
   plot(index, s);
end

xlabel('Neuron index');
ylabel('Signal strength');
title('(a)');

subplot(1,2,2);
hold on

feedback = 0.75;
s = zeros(50);

for t=1:15
   for i =1:50
      activation(i) = input(i);
      for j=1:50
         activation(i) = activation(i) + feedback*w(j,i)*s(j);
      end
   end
   
   for i=1:50
      if (activation(i) > max) s(i) = max;
      elseif (activation(i) < 0) s(i) = 0;
      else s(i) = activation(i);
      end
   end
   
   plot(index, s);
end
axis([0 50 0 5]);
xlabel('Neuron index');
ylabel('Signal strength');
title('(b)');