| | |
Summary: Supernova Recognition using Support Vector Machines
Raquel A. Romano Cecilia R. Aragon Chris Ding
Computational Research Division
Lawrence Berkeley National Laboratory
1 Cyclotron Road, Berkeley, CA 94720
E-mail: {romano,aragon,chqding}@hpcrd.lbl.gov
Abstract
We introduce a novel application of Support Vector Ma-
chines (SVMs) to the problem of identifying potential su-
pernovae using photometric and geometric features com-
puted from astronomical imagery. The challenges of this
supervised learning application are significant: 1) noisy
and corrupt imagery resulting in high levels of feature un-
certainty, 2) features with heavy-tailed, peaked distribu-
tions, 3) extremely imbalanced and overlapping positive
and negative data sets, and 4) the need to reach high posi-
tive classification rates, i.e. to find all potential supernovae,
while reducing the burdensome workload of manually ex-
amining false positives. High accuracy is achieved via a
sign-preserving, shifted log transform applied to features
|
