Mathematics – Probability
Scientific paper
Dec 2006
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2006aas...209.7802v&link_type=abstract
2007 AAS/AAPT Joint Meeting, American Astronomical Society Meeting 209, #78.02; Bulletin of the American Astronomical Society, V
Mathematics
Probability
Scientific paper
Image metadata describes the properties of an image and can be used for classification, e.g., galactic, extra-galactic, solar system, standard star, among others. We are developing a data mining application to automate such a classification process based on supervised learning using decision trees. We are applying this application to the NOAO NVO Portal (www.nvo.noao.edu). The core concepts of Quinlan's C4.5 decision tree induction algorithm are used to train, build a decision tree, and generate classification rules. These rules are then used to classify previously unseen image metadata. We utilize a collection of decision trees instead of a single classifier and average the classification probabilities. The concept of ``Bagging'' was used to create the collection of classifiers. The classification algorithm also facilitates the addition of weights to the probability estimate of the classes when prior knowledge of the class distribution is known.
Barg Irene
Miller Christopher J.
Smith Christopher R.
Vaswani Pooja
No associations
LandOfFree
Datamining the NOAO NVO Portal: Automated Image Classification does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Datamining the NOAO NVO Portal: Automated Image Classification, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Datamining the NOAO NVO Portal: Automated Image Classification will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1158036