Search All of the Math Forum:
Views expressed in these public forums are not endorsed by
Drexel University or The Math Forum.
|
|
|
|
Re: Hopfield Neural Network
Posted:
Feb 20, 2013 1:48 PM
|
|
"Prateek " <malikprateek22@gmail.com> wrote in message
news:kg336s$fph$1@newscl01ah.mathworks.com...
> "Steven_Lord" <slord@mathworks.com> wrote in message
> <kg2uul$r6n$1@newscl01ah.mathworks.com>...
*snip*
> hello Steve i actually have 26 images of size 100 X 100 stored as a vector
> each having 10000 elements each so we make a matrix of 26 x 10000 and i
> want to store each image on to the network so let me know how should i try
> with transpose or with out it because when i tried it without transpose it
> actually works but then when i try to test with a distorted image problems
> crop up
To paraphrase Roy Scheider, I think you're gonna need a bigger data set.[*]
I suspect trying to determine how the 10000 pixels interrelate from just 26
examples will require a lot of extrapolation and/or guessing or will require
the pixels to be very highly correlated with one another.
--
Steve Lord
slord@mathworks.com
To contact Technical Support use the Contact Us link on
http://www.mathworks.com
[*] The original was about a boat and IIRC is often misquoted (we're instead
of you're.) http://en.wikiquote.org/wiki/Jaws_(film)#Chief_Martin_Brody
|
|
|
|
