High error percentage after training neural network

Hello,

I am working on Handwritten character recognition .I have created feature vector of size 130*85 (since there are 5 samples for each character) and a target vector of size 130*26 (26 classes). But when I feed these to neural network , I am getting a high error percentage. Please tell me whether I am proceeding correctly.

Please find attached the input vector and the target vector files.

I have extracted the features using this code http://www.mathworks.in/matlabcentral/fileexchange/24624-feature-extraction-for-character-recognition

Answer 1

[Greg Heath]

 [ 85 130 ] = size(input)

 [ 26 130 ] = size(target)

 What non-default input parameters did you use?