Prev: how to run a .nc file in matlab
Next: Statistics on nonequally spaced data in 2D
From: saranya on 10 Mar 2010 06:53
Hello,
how to train the dataset (.mat format) .i.e connected to matlab from oracle using odbc. could anyone please tell how to train dataset since im new to matlab.
From: wyomesh on 10 Mar 2010 23:07
On Mar 10, 4:53 pm, "saranya " <rpsara...(a)gmail.com> wrote:
> Hello,
> how to train the dataset (.mat format) .i.e connected to matlab from oracle using odbc. could anyone please tell how to train dataset since im new to matlab.

hello

you import first the dataset in matlab in the form of a matrix. then
make a TARGET MATRIX if your network requires one. SOM (Self Organised
Map) does not necessarily require a TARGET in training while a FEED
FORWARD NETWORK requires both the INPUT MATRIX and TARGET MATRIX. For
early stopping of network you will have to customise some network
parameters and you may have to submit VALIDATION Data. MATLAB can
automatically divide your data into training set, test set and
validation set. see dividerand etc. so training set is a small set of
data used in training. test data is used for simulation of network and
MATLAB uses validation data to early stop the network when validation
errors increased by a particular amount.

so, there are so many things u have to look on simultaneously.

have fun bye
From: saranya on 11 Mar 2010 00:01
hello,
Thank u so much for ur reply..im not using neural networks..im gonna use it in fuzzy logic..i need to know, how to train the dataset.could u please help of it.
 | 
Pages: 1
Prev: how to run a .nc file in matlab
Next: Statistics on nonequally spaced data in 2D