Archived

This topic is now archived and is closed to further replies.

Dark Star

NNs and Backpropagation

Recommended Posts

Hi Guys, I was wondering with back prop, the error values u calculate from the formula:
 error[i] = desired[i] - output[i];   (Im aware in some   
                                       literature its opposite)
does this error term get backpropated to only the previous layer of neurons or can it be used for the layer after or does a new error term have to be calculated and applied to the next layer back? Also consider this network structure below
I        1           2         O

o                              o 
         o                     o
                    o          o 
o        o                     o
                    o          o
o        o                     o
                               o
o                              o
Layers I and O are input and output layers respectively and 1 and 2 are hidden layers. ''o'' is a neuron. This is a forward feed NN. There are 8 outputs so there would be 8 error terms in the error used for backpropagation. right? So how will I use 8 errors to adjust the weights in layer 2? Layer 2 has two neurons and would have 6 weights all together (I think!) How would I used 8 error terms to adjust 6 weights? And let me ask first question again. When I adjusted the weights in layer 2 can I use that error term to adjust the weights in Layer 1 or do the have to be calculated again? Do the output neurons require any weights from Layer 2? Finally, do the input and output layers have any activation functions for their neurons. So before a neuron in the input layer fires and output to the first hidden layer, must it undergo an activation function or does it just pass its input to the layer 1 and before we extract the outputs in the output layer must it undergo an activation function or do we just get the output? Sorry for all those questions, its a bad habit to ask soo much but its for a university project and I pressed for time even though the semester aint started yet. Thanks in advance. DarkStar UK o [edited by - dark star on January 4, 2004 3:19:38 PM] ------------------------------- Loves cross-posting because it works

Share this post


Link to post
Share on other sites
Guest Anonymous Poster
Syntax, these slides are great, do you have other pdf from your professor (about NN or other AI related topics) ?

Share this post


Link to post
Share on other sites
quote:
Original post by Anonymous Poster
Syntax, these slides are great, do you have other pdf from your professor (about NN or other AI related topics) ?



Here''s a tip: when searching online (at Google, Dogpile, whatever), include search terms like "course notes", "lecture notes", "course readings", etc. to steer the search towards material provided by professors'' for college and university courses. Such material is usually well-written, well-formatted and often quite accesible ("readable"), without the boring Ph.D. dissertation blather or HTML junk so common on the Web.

-Predictor
http://will.dwinnell.com


Share this post


Link to post
Share on other sites
yes i do, infact it is at the same site
It has a really good one for Adaptive Resonance Theory (ART1)
If patter learning is your thing study (ART1 & ART2) those networks

www.jodypaul.com/cs/ai

-lucas

Share this post


Link to post
Share on other sites
I should point out one important thing (that may or may not be relevant in this case)...

Be VERY careful about posting lecture or course notes on the web. If you haven''t sought the permission of the author to do so, then you are in breach of their copyright.

If, however, the notes are placed on the web by the author themself, then its okay to link to them (in most cases).

Cheers,

Timkin

Share this post


Link to post
Share on other sites

The courses and lecture notes provided are released on the web, publicaly, with respective copyrights inplace. I am sure my professor will have no qualms with the notes being released. If it is the will of Gamedev.net administrators, I would be more than happy to get official permission for the release. But it is not needed.

-Lucas

Share this post


Link to post
Share on other sites
quote:
Original post by Syntax

The courses and lecture notes provided are released on the web, publicaly, with respective copyrights inplace. I am sure my professor will have no qualms with the notes being released. If it is the will of Gamedev.net administrators, I would be more than happy to get official permission for the release. But it is not needed.

-Lucas


That wasn''t the point of my post Syntax... (read the first line again)... My post was a general warning to everyone reading the thread of the dangers of posting information on the web that wasn''t authored by the poster. I''d hate to see someone get into trouble with one of their professors when they thought they were being helpful!

Cheers,

Timkin

Share this post


Link to post
Share on other sites