| |
 |
| Paper: |
Algorithmic and Machine Learning Approaches to Automatic Identification of Peculiar Galaxies in Large Astronomical Databases |
| Volume: |
538, ADASS XXXII |
| Page: |
293 |
| Authors: |
Lior Shamir |
| DOI: |
10.26624/PTTV2181 |
| Abstract: |
Digital sky surveys can image many millions of extra-galactic objects.
While most of these objects are galaxies of known types, a small portion have rarely
or never been seen before. These objects possess critical information about the past,
present, or future Universe. Since they are hidden inside very large databases, finding
them is impractical by manual analysis, and therefore, automation is required. Different approaches for automatic identification include model-driven and data-driven approaches, which can be further separated into supervised and unsupervised machine
learning. Due to the extreme size of these databases, even a mild rate of false positives
will make the output unmanageable. Therefore, a practical solution to detect peculiar
objects must be able to control the false-positive rate while also handling artifacts or saturated images, which are common in digital sky surveys. The algorithms, approaches,
and examples of detected objects are described, with application to data from DES, SDSS, and HST. |
|
|
|
|
|
