PARALLEL DATA LAB

DISC-QUASAR CODE DISTRIBUTION:

Identifying Distant Quasars in Sky Surveys

Data-mining tools for identifying quasars in sky-survey datasets based on their brightness in the five standard passbands, that is, in the images made through five different color filters.

The DISC-Quasar package is a set of tools for distinguishing quasars from other astronomical objects, such as stars and galaxies. Its main focus is identifying distant quasars, with redshift greater than 2.3, which means that they are over twelve billion light years away. The package includes supervised learning algorithms that construct rules for quasar detection based on labeled samples.

   download
   ZIP ARCHIVE

MAIN FEATURES

  • Application of support vector machines, decision trees, and nearest-neighbor algorithms.
  • Identifying distant quasars with 81% precision, which means that 81% of the objects selected by the system are true quasars.

LIMITATIONS

  • This work is still in its early stage, and the released algorithms are preliminary and quite messy.
  • The recall of the developed techniques is low. It is about 30%, which means that the system detects only 30% of distant quasars and misses the other 70%.
  • The current system does not include a distributed version and does not scale to datasets with more than a few million objects.

Please review our software license.