Robust Machine Learning Applied to Terascale Astronomical Datasets - AstrophysicsReportar como inadecuado

Robust Machine Learning Applied to Terascale Astronomical Datasets - Astrophysics - Descarga este documento en PDF. Documentación en PDF para descargar gratis. Disponible también para leer online.

Abstract: We present recent results from the Laboratory for Cosmological Data Miningthis http URL at the National Center for SupercomputingApplications NCSA to provide robust classifications and photometric redshiftsfor objects in the terascale-class Sloan Digital Sky Survey SDSS. Through acombination of machine learning in the form of decision trees, k-nearestneighbor, and genetic algorithms, the use of supercomputing resources at NCSA,and the cyberenvironment Data-to-Knowledge, we are able to provide improvedclassifications for over 100 million objects in the SDSS, improved photometricredshifts, and a full exploitation of the powerful k-nearest neighboralgorithm. This work is the first to apply the full power of these algorithmsto contemporary terascale astronomical datasets, and the improvement overexisting results is demonstrable. We discuss issues that we have encountered indealing with data on the terascale, and possible solutions that can beimplemented to deal with upcoming petascale datasets.

Autor: Nicholas M. Ball, Robert J. Brunner, Adam D. Myers University of Illinois at Urbana-Champaign, National Center for Supercomputing


Documentos relacionados