Summer research project at Mount Allison University.
Monday, May 30, 2011
ANN (Works only for the testing file I used?)
Cleaned up the testing data to be much more simplified, and now the ANN won't use it. The format of the file is entirely identical, and everything gets set up correctly but it won't run the learner.
For the input format, did you reduce the number of neuron in the output layer of the ANN? The only difference between the sets should be the reduced number of classes. Also, to make a simple test set, I'd reduce the number of test samples for each genera to be equal. Right now the dark blue one still dominates the population and will likely skew the classifier.
For the charts, circularity looks like a good feature. Is circularity just the ratio of major and minor axis? I didn't see any clear clustering for Ferat's diameter. There may have been a little bit of clustering with area (the burgundy class had most of the large areas). Did you try a ratio of area to perimeter? A unit-less ratio may be more useful.
For the input format, did you reduce the number of neuron in the output layer of the ANN? The only difference between the sets should be the reduced number of classes. Also, to make a simple test set, I'd reduce the number of test samples for each genera to be equal. Right now the dark blue one still dominates the population and will likely skew the classifier.
ReplyDeleteFor the charts, circularity looks like a good feature. Is circularity just the ratio of major and minor axis? I didn't see any clear clustering for Ferat's diameter. There may have been a little bit of clustering with area (the burgundy class had most of the large areas). Did you try a ratio of area to perimeter? A unit-less ratio may be more useful.
I'm going to bring in the Cintiq tablet tomorrow.