Set initial weights of neural networks

Hello everyone !

I would like to set the initial weights of artificial neural network in order to define relevant number of hidden nodes. So, I wrote the code below but it does not work. Could you correct it for me? Thank you.

% Solve an Input-Output Fitting problem with a Neural Network

function [performance] = NN(x,t,i)

% Choose a Training Function
% For a list of all training functions type: help nntrain
% 'trainlm' is usually fastest.
% 'trainbr' takes longer but may be better for challenging problems.
% 'trainscg' uses less memory. NFTOOL falls back to this in low memory situations.
[xn,xs] = mapminmax(x);% Normalization of inputs
[tn,ts] = mapminmax(t); % Normalization of outputs
%dx = randperm(size(xn,2)); % Inputs permutation
%xn = xn(:,dx); % Inputs permutation
%dx = randperm(size(tn,2)); %  Output normalization
%tn = tn(:,dx); % Output normalization

trainFcn = 'trainlm';  % Levenberg-Marquardt
% Create a Fitting Network
H = i ;
net = fitnet(H, trainFcn);

I = size(x,1);
N = size(x,2);
O = size(t,1);

net = configure(net, xn, tn);

% Define initial conditions (transferFn, weights, biases...)

% Define transfer Function for each layer
net.layers{1}.transferFcn = 'tansig';
net.layers{2}.transferFcn = 'logsig';

rng(0);
IW = 0.001 * randn(H,I);
LW = 0.001 * randn(O,H);
b1 = 0.001 * randn(H,1);
b2 = 0.001 * randn(O,1);

% Show weights and biases before training
net.IW{1,1}
net.LW{2,1}
net.b{1}
net.b{2}

% Setup Division of Data for Training, Validation, Testing
%net.divideParam.trainRatio = 70/100;
%net.divideParam.valRatio = 15/100;
%net.divideParam.testRatio = 15/100;

net.divideFcn = 'dividerand';

% Train the Network
[net] = train(net,xn,tn);

% Show weights and biases after training
net.IW{1,1} = IW;
net.LW{2,1} = LW;
net.b{1} = b1;
net.b{2} = b2;

% Test the Network
y = net(xn);
e = gsubtract(tn,y);
performance = perform(net,tn,y)

[Santhana Raj]

What do you mean by not working???

If you are getting error, then post the error.

If it is not able to train a specific data, then post the data also.

If you are expecting some results and the results are different, then post the data along with the expected results.