On 7/26/06, Dinesh Joshi <dinesh.a.joshi(a)gmail.com> wrote:
Hi list,
I am currently learning neural networks as a hobby. I would like to
implement them in C and on Linux. I have googled around and there seems
to be immense data on them yet its all theoretical. I am currently
learning the BPN algorithm whose logic I am able to understand but the
math is a bit tough. I would like to know if there is/are some book /
website / white papers which talk about Neural Networks from
implementation / performance point of view. Oh yes, it would be great
help if the math is explained in simpler terms :)
Well, a long time before I was working on neural networks. A good amount
of
explanation is given in the MATLAB manual(i mean the real basics are
explained..).You can get the manual free(as in beer) over the web. In linux
you have Scilab which comes with an ANN tolbox. Scilab is not GPLed...comes
with its own scilab license. (check octave also)
For the mathematics part of it..u can start with this:
http://www.mth.kcl.ac.uk/courses/guide.pdf..
I think it would be wiser to use the libraries already available(unless u
want to reinvent the wheel!!)(checkout fann, libeneural)
Rgds
Harish
P.S: This book was told by my prof. as a good one to start with: "Neural
Networks Algorithms, Applications, and Programming Techniques" by James A.
Freeman and David M. Skapura (Addison-Wesley, 1991).