Massively Parallel Machine Learning in the Virtual Observatory as a Key Technology in the Era of Multi-Million Spectral Surveys

0562031 - ASÚ 2023 RIV IN eng C - Konferenční příspěvek (zahraniční konf.)
Škoda, Petr
Massively Parallel Machine Learning in the Virtual Observatory as a Key Technology in the Era of Multi-Million Spectral Surveys.
Proceedings of the International Workshop on Stellar Spectral Libraries: IWSSL 2017. Calcutta: Astronomical Society of India, 2017, s. 73-82. ASI Conference Series, 14. ISBN 978-81-935285-1-8.
[Proceedings of the International Workshop on Stellar Spectral Libraries. Sao Paulo (BR), 06.02.2017-10.02.2017]
Institucionální podpora: RVO:67985815
Klíčová slova: Be stars * emission-line * data analysis
Obor OECD: Astronomy (including astrophysics,space science)
https://articles.adsabs.harvard.edu/pdf/2017ASInC..14...73S

In this paper we propose a new approach that uses modern machine-learning techniques as semi-supervised training, deep learning, or outlier detecting that helps to identify specific rare cases of unusual objects like stars with strong emission lines or P-Cyg profiles, or blazars, as well as to eliminate the instrumental and processing artefacts which cannot be handled correctly by a normal streaming pipeline. The amount of data and time-absorbing algorithms require a `Big Data' approach, using massively parallel processing in the cloud by applying modern technologies such as GPGPUs, Hadoop and Spark.
Trvalý link: https://hdl.handle.net/11104/0334454