Classifying the high energy universe with ClassX
Thomas A. McGlynn, Anatoly A. Suchkov, E. L. Winter, L. Angelini, Michael F. Corcoran, S. Derrière, Megan E. Donahue, S. A. Drake, P. Fernique, Francoise Genova, R. J. Hanisch, F. Ochsenbein, W. D. Pence, Marc Postman, Nicholas E. White, Richard L. White · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2002
Building an automated classifier for high-energy sources provides an opportunity to prototype approaches to building the Virtual Observatory with a substantial immediate scientific return. The ClassX collaboration is combining existing data resources with trainable classifiers to build a tool that classifies lists of objects presented to it. In our first year the collaboration has concentrated on developing pipeline software that finds and combines information of interest and in exploring the issues that will be needed for successful classification. ClassX must deal with many key VO issues: automating access to remote data resources, combining heterogeneous data and dealing with large data volumes. While the VO must attempt to deal with these problems in a generic way, the clear science goals of ClassX allow us to act as a pathfinder exploring particular approaches to addressing these issues.