>tfw too stupid too understand Machine Learning
feels bad man
For example i look at this example:
https://medium.com/coinmonks/implementing-an-artificial-neural-network-in-pure-java-no-external-dependencies-975749a38114
But i dont know why there are such calculations, for example why do i use Weight2 in
double[][] dZ1 = multiply(dot(T(W2), dZ2), subtract(1.0, power(A1, 2)));
and why do i need to multiply by 1-A1?
I try to modify this code, so i can predict fifth set of continuous values by first four sets, like that:
X={{2,2,4},{5,2,5},{2,5,6},{5,2,4}}
Y={{2},{4},{5},{7}}
but i got wrong dimensions in some place, because i get out of bounds of array while calculating cross product. I try to draw this example on paper so i see exactly what are dimmensions of matrix inputs and matrix outputs in certain function calls, but its so cumbersome