Neural Nets

Go To Last Post
6 posts / 0 new
Author
Message
#1
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Got two books on Neural Nets, neither cover noting more than 1 hidden layer.

 

Oh, anyone recombed a book, really would like one that covered two or more hidden layers..

This topic has a solution.
Last Edited: Sat. May 9, 2020 - 09:50 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Anyone here good with Neural Nets or equally really good at maths.  Have a look at the attached document with the full question listed!

 

Attachment(s): 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

try this...has code you can try & you can instal Python for free  (get Python at  https://www.python.org/)

 

https://machinelearningmastery.com/implement-backpropagation-algorithm-scratch-python/

 

Another decent read:
https://medium.com/datathings/neural-networks-and-backpropagation-explained-in-a-simple-way-f540a3611f5e

 

I  somewhat remember a course I took around 1993 using my 100 MHz PC to train on Lenna:

https://en.wikipedia.org/wiki/Lenna

 

 

When in the dark remember-the future looks brighter than ever.   I look forward to being able to predict the future!

Last Edited: Sun. Apr 19, 2020 - 03:59 PM
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0


Hello freaks.  So I was wondering if you could help me?  I'm looking to know how to calculate a 2 hidden layer network.  Have a look at the diagram below.

To calculate w1 I have the following.

\frac{dE}{dw_1} = \frac{dE}{dh_1} \frac{dh_1}{dz_{h_1}} \frac{dz_{h_1}}{dw_1}

The calculation of the first term on the right hand side of the equation above is a bit more involved than previous calculations since h_1 affects the error through both o_1 and o_2.

\frac{dE}{dh_1} = \frac{dE}{do_1} \frac{do_1}{dz_{o_1}} \frac{dz_{o_1}}{dh_1} + \frac{dE}{do_2} \frac{do_2}{dz_{o_2}} \frac{dz_{o_2}}{dh_1}

 

Now my question is.  If I had a hypothetical weight input into X1 called weight I, wi what is the equation for back propagating tht weight? I'm thinking it is maybe it is this.

∂E/∂wi = ∂E/∂x1  .  ∂x1/∂zx1  .  ∂zx1/∂w1

 

Therefore:

 âˆ‚E/∂x1 = ( ∂E/∂o1 .  ∂o1/∂zo1  .  ∂zo1/∂x1)  +  (∂E/∂o2  .  ∂o2/∂zo2  .  ∂zo2/∂x1)

 

Or maybe it is this:

 âˆ‚E/∂x1 = ( ∂E/∂h1 .  ∂h1/∂zh1  .  ∂zh1/∂x1)  +  (∂E/∂h2  .  ∂h2/∂zh2  .  ∂zh2/∂x1)

 

I really hope you can help me with this, thanks for having a look!

 

 

 

 

Wm.

 

  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

Ah, I didn't see this thread before I replied to your other one here:

https://www.avrfreaks.net/comment/2896386#comment-2896386

 

... so never mind ;-)

"Experience is what enables you to recognise a mistake the second time you make it."

"Good judgement comes from experience.  Experience comes from bad judgement."

"Wisdom is always wont to arrive late, and to be a little approximate on first possession."

"When you hear hoofbeats, think horses, not unicorns."

"Fast.  Cheap.  Good.  Pick two."

"We see a lot of arses on handlebars around here." - [J Ekdahl]

 

This reply has been marked as the solution. 
  • 1
  • 2
  • 3
  • 4
  • 5
Total votes: 0

This function does it, just substitute for db1 for the weights.

 

\frac{dE}{db_1} = \frac{dE}{do_1} \frac{do_1}{dz_{o_1}} \frac{dz_{o_1}}{dh_1} \frac{dh_1}{dz_{h_1}} \frac{dz_{h_1}}{db_1} + \frac{dE}{do_2} \frac{do_2}{dz_{o_2}} \frac{dz_{o_2}}{dh_2} \frac{dh_2}{dz_{h_2}} \frac{dz_{h_2}}{db_1}