bug-gnubg
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [Bug-gnubg] Neural Net Tutorials


From: OJOHANS
Subject: RE: [Bug-gnubg] Neural Net Tutorials
Date: Mon, 9 Mar 2009 09:48:34 +0100

Hi again,

Sorry the delay. The weekends are .... (yeah, you know..)

>I did some calculus at school, but that was years ago. I 
>haven't studied either of the other two though. I'm willing to 
>do some learning if these subjects are prerequisites for a 
>good understanding of neural networks.

I think I would start with simple linear algebra. Just find a really basic book 
on linear algebra and learn the basics. Learn the notation and learn basic 
matrix operations like multiplying, transposing and inverting. If you know 
those things it's much simpler to explain the neural net algorithms.

Run to your library and get a book on linear algebra, any book will do. Then 
learn the basics of linear algebra. Anton or Lay are good books, or maybe one 
of those popular titles including words like "demystified", "dummies" or "24 
hours".  Gilbert Strang is also fine, but it way over what you need to get a 
basic understanding of neural networks.

Matrix basics:
- Adding/subtracting
- Multiplying
- Transposing
- Determinants
- Inverting
- Identity matrices
- Basic matrix expressions

Things you _do_not_ need:
- Eigenvector/eigenvalues
- SVD
- Hermittian matrices
- Different decompositions (QR, Schur, Gram Schmidt etc).
(stop reading before you reach any of these subjects)

While you're at the library, try to find:
"Understanding Neural Networks And Fuzzy Logic", by Kartalopous, ISBN:8120316800

It's small and well written, does not include hairy details, and give a nice 
overview. (However, you need the basic linear algebra). Tell me if you find 
this book and I can point out the important chapters.

Next question: Are you a coder? Do you code anything? Do you want to learn 
coding? Any simple experience with any programming language? Pascal in 
pre-school?
 
-Øystein


-------------------------------------------------------------------
The information contained in this message may be CONFIDENTIAL and is
intended for the addressee only. Any unauthorised use, dissemination of the
information or copying of this message is prohibited. If you are not the
addressee, please notify the sender immediately by return e-mail and delete
this message.
Thank you.




reply via email to

[Prev in Thread] Current Thread [Next in Thread]