3

I am trying to make my own neural network in order to understand how it works, I do understand some basic differential calculus, but the back propagation formulas are a bit hard for my brain.

This is as far as I could go:

http://badjano.com/nn/stackoverflow.html

I wanted it to work with multiple hidden layers ( only work with 1 though with as many neurons as I want ) and also the xor test isn't working either and so is non-linear tests, and I don't get why it isn't working.

I have some neural network codes that I found over the internet, but I had some difficulty trying to read it.

If anyone could help, I would very much appreciate it :)

Edit: I couldn't find an easier way to show my code, but I did put it all in the html file so all you need to do is view the source, or save it locally so you could edit it.

Badjano
  • 418
  • 7
  • 23
  • This is more of the question might be better for the [Computer Science](http://cs.stackexchange.com/) or [Statistics](http://stats.stackexchange.com/tour) stack exchange sites, note it's currently too broad. Are you sure your properly propagating through each layer? There are several ways to do it. – Spencer Wieczorek Jul 12 '16 at 03:39
  • As far as I checked the propagation is going through the layers, but I'm not sure if the algorithm is completely right, I collected some information I found but I don't know if I applied it correctly. Thanks, I will try and post on those stacks. – Badjano Jul 12 '16 at 18:04
  • I see you tagged Javascript, maybe check out some JS neural network libraries like [Neataptic](https://github.com/wagenaartje/neataptic) to see how they do it. – Thomas Wagenaar Apr 19 '17 at 11:49

0 Answers0