Code covered by the BSD License  

Highlights from
Training MLP by means of Cuckoo Optimization Algorithm (COA)

image thumbnail
from Training MLP by means of Cuckoo Optimization Algorithm (COA) by Ramin Rajabioun
MultiLayer Perceptron trained by Cuckoo optimization Algorithm.

ANN_Cost(POP) 
function cost = ANN_Cost(POP)

global input output numOfNeurons NumofInputs

cost = zeros(size(POP,1),1);

for ipop = 1:size(POP,1)
    pop = POP(ipop,:);

    W = pop(1:(NumofInputs+1)*numOfNeurons);
    U = pop((NumofInputs+1)*numOfNeurons+1 : end);

    z1 = tansig(input*W(1)+W(4));
    z2 = tansig(input*W(2)+W(5));
    z3 = tansig(input*W(3)+W(6));
    z4 = 1;   % bias
    
    ANN_output = z1*U(1) + z2*U(2) + z3*U(3) + z4*U(4);
    
    cost(ipop) = sum( ( (output - ANN_output).^2 ) )./ sum(output.^2) * 100;
    
end

Contact us