Monday, September 24, 2012
Astronomy Tea Talk
How to Find the Rare Sources All New Surveys Promise: Machine-learning Enabled Classification and Discovery
Adam Miller, UC Berkeley
The falling costs of computing and CCD detectors has led to a great boom in wide-field time-domain optical surveys during the past ~decade, with several new surveys expected prior to the arrival of LSST. This observational boon, however, comes with a catch: the data rates from these surveys are so large that discovery techniques heavily dependent on human intervention are becoming unviable. Identifying the extraordinary within large data sets requires an understanding and model of the more common sources. In this talk I will detail new methods, which utilize semi-supervised machine-learning algorithms, to automatically classify the light curves of time-variable sources. Using these methods, we have produced a data-driven probabilistic catalog of variables found in the All Sky Automated Survey (ASAS). We identify true outliers within the data via the clustering of sources which are not classified with high significance. This has led to the discovery of several rare systems: including four new members of the R Coronae Borealis class, which are thought to be the product of low mass white dwarf mergers, but are underrepresented in the Milky Way relative to the LMC. Finally, I will describe how a machine-learned robot discovered two outbursting young stars which are changing the way we understand star formation.