Deep-Learning the Time Domain. Issue Volume 14:Issue S339(2018) (November 2017)

• Record Type: Journal Article
• Authors: Mahabal, A.; Sheth, K.; Gieseke, F.; Drake, A.; Djorgovski, G.; Graham, M. J.
• Editors: Griffin, E.
• Is Part Of: Proceedings of the International Astronomical Union. Volume 14:Issue S339(2018)
• Page Start: 165
• Page End: 171
• Publication Date: 2017-11
• Subjects: Surveys, -- methods: data analysis, -- Techniques: image processing; Astronomy -- Congresses; Astronomy -- Periodicals; 520
• DOI: 10.1017/S1743921318002491
• Languages: English
• ISSNs: 1743-9213

Abstract

Abstract: "Deep learning" is finding more and more applications everywhere, and astronomy is not an exception. This talk described the application of convolutional neural networks to time-domain astronomy, specifically to light-curves of sources. The work that is discussed is based on a published paper to which reference can be made for more detail. The talk finished with a note cautioning new practitioners about the pitfalls lurking in out-of-the-box use of deep-learning techniques.