avaisp / machine-learning-programming-assignments-coursera-andrew-ng Goto Github PK

I am confused that what this line actually doing
y = repmat([1:num_labels], m, 1) == repmat(y, 1, num_labels);
Please explain it to me.
I've printed this y and got 5000*10 matrix but why this is necessary. and please tell me how this command works.

??? Input argument "X" is undefined.

Error in ==> featureNormalize at 9
X_norm = X;

how to run Logistic Regression Cost using fminunc....where to write fminunc commands in code.

I noticed the following issues:

1. You are using dot product which won't output a single-digit number but a matrix.
2. using y without taking its transpose will lead to errors.

Replace the code with the following to fix it. (Using cross product instead of dot product).
J = (1/m) * (-y' * log(sigmoid(X * theta)) - (1 - y)' * log(1 - sigmoid(X * theta)));

for e.g. i have five layers:
first layer 30,2nd l 25,3rd l 25,4th l 10,5th 8000
during backpropagation delta value calculations i have d5=(11800) so when i calculate d4 i have d4 of size(111) including bias. but for d3 my theta is 2610but d4 is 111 and thats the problem. My question is should i include bias as well

the code in the function block is not the complete file l

https://github.com/AvaisP/machine-learning-programming-assignments-coursera-andrew-ng/blob/master/machine-learning-ex7/ex7/kMeansInitCentroids.m