<div class="gmail_quote">On Mon, Jun 15, 2009 at 5:00 PM, Trin Trin <span dir="ltr"><<a href="http://trin.cz">trin.cz</a>@<a href="http://gmail.com">gmail.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div>Hi Alp,<br></div><div></div><div>- even with correctly programmed back-propagation, it is usually hard to make the net converge. </div></blockquote><div><br>Yeah, I know, that's why we're training it until the quadratic error goes under 0.1.<br>
<br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div></div><div>- usually you initialize neuron weights with somewhat random values, when working with back-propagation.</div>
</blockquote><div><br>Yeah, that'll be done too, once the algorithm will be ready. I'll provide fancy and easy functions to create a neural net just giving the numbers of layers and their sizes.<br> <br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div></div>
<div>- do some debug prints of the net error while training to see how it is going</div></blockquote><div><br>Good idea, yeah.<br> <br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<div></div><div>- xor function cannot be trained with a single layer neural net !!!</div></blockquote><div><br>That's why there are 2 layers there, one hidden and the output one. I consider the "inputs" as ... inputs, not as a first layer of the NN.<br>
</div></div><br>Thanls for your time. If you have any clues when reading the code, don't hesitate of course.<br><br>-- <br>Alp Mestan<br><a href="http://blog.mestan.fr/">http://blog.mestan.fr/</a><br><a href="http://alp.developpez.com/">http://alp.developpez.com/</a><br>