Prev: Open Office Spreadsheet
Next: data fitting for 2 data sets and functions
From: chris Austin on 20 Feb 2010 14:20
Hello,

I managed (can't code well in matlab) to code a simple feedfoward Neural net and each time i ran the code i get the following error.

"??? Index exceeds matrix dimensions.
Error in ==> train>trainLayer at 35
momentum(i, size(input)) = eta * bias * deltas(i) + alpha * momentum(i, size(input)
+ 1);
Error in ==> train at 23
[outputWeights, oMomentum] = trainLayer(hiddenOutput, outputWeights, oMomentum,
outputDeltas);

Error in ==> test at 14
train(img, hWeights, oWeights, target, hMom, oMom);"


here (second to the las line of the code) is the portion of the code the error is pointing to.... What is possible wrong with it?


function [weights, momentum] = trainLayer(input, weights, momentum, deltas)
eta = 0.1;
alpha = 0.9;
bias = 1;

for i = 1 : size(weights)
for j = 1 : size(input)
momentum(i, j) = eta * input(j) * deltas(i) + alpha * momentum(i, j);
weights(i, j) = weights(i, j) + momentum(i, j);
end
momentum(i, size(input)) = eta * bias * deltas(i) + alpha * momentum(i, size(input) + 1);
weights(i, size(input)) = weights(i, size(input)) + momentum(i, size(input) + 1);
end
 | 
Pages: 1
Prev: Open Office Spreadsheet
Next: data fitting for 2 data sets and functions