neural network as classifier

[kamble priyanka]

hello,
i am working on leaf classification. and i am using neural network as classifier for best result. but i don't know to how to give input to neural network classifier.
plz give me suggetion.

 

[emresel]

Hi;
First you need to generate/extract some information from your images (here leaves i think). It is called feature extraction. You shouldn't supply raw data to NN. Otherwise it would be cumbersome.

Good luck

 

[kamble priyanka]

hi,
i have done extraction. but now my task is give that input to neural network. what can i do in that case?

 

[cherryman]

Probably You have to construct 2 matrices. You should have several examples (of Your leaves) to learn Your network, so those examples will be placed in rows. Each example will have features that You extracted earlier. Features will be placed in columns. As a result You should have:

input = [F11, F21 ,F31, ..., Fm1;
           ...;
           F1n, F2n ,F3n, ..., Fmn]

Where n - no of examples and m - no of features.
Second matrix is a class matrix (target matrix). It can be binary matrix 1 x N if you have only two classes to detect.
But in fact the process of providing data is depended on the tool that you use.

 

[emresel]

Number of your features will define the number of input neurons of your NN. Number of hidden layers and hidden neurons depends on your experiments. At first step one of the important thing is to normalize your features. ie your features let's say length (in mm) and number of major veins on leaf. These are in different spaces and since lenghts in mm may have greater numerical effect on the NN. Thats result is biasing your network unintentionally. So before you apply feature set to your NN, you should normalize and make the units and their strength (magnitudes) pointless.
As i remember there is good example in MATLAB help or demos, that shows that normalizing steps...
Good luck