How to calculate accracy of neural network using ANN toolbox matlab
2 views (last 30 days)
Show older comments
Hi all can anyone tell me how do i find the accuracy measure of neural network when using nntoolbox for pattern recognition? If I get mse on trainning= 2.63617e-7 and %E=0; mse on validation= 1.00120e-2 and %E=6.666e-0; and mse on testing=3.73162e-3 %E=0, how can i calculate the accuracy??? and how can i know if my network is overfitting. Thanks very much!
0 Comments
Accepted Answer
Greg Heath
on 7 Mar 2012
Number of classes c = ?
Number of input/target examples per class Ni = ? % i= 1:c
Dimensionality of input I = ?
Dimensionality of output O = ? % c or c-1?
Total number of I/O pairs N = sum(Ni) = ?
Data division ratio per class Ntrni/Nvali/Ntrni = ? % i= 1:c
Number of hidden nodes H = ?
Number of random weight initialization trials Ntrials = ?
Min, median, mean, std and max of PctErr (%E invalid name) over Ntrials
Similar tabulations per class.
Although training goal is to minimize MSEval because PctErrval is discontinuous, the only important results are the PctErrval and PctErrtst stats.
Hope this helps.
Greg
0 Comments
More Answers (0)
See Also
Categories
Find more on Sequence and Numeric Feature Data Workflows in Help Center and File Exchange
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!