BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.adass2021.ac.za//8XXCYS
BEGIN:VTIMEZONE
TZID:Africa/Johannesburg
BEGIN:STANDARD
DTSTART:20000101T000000
RRULE:FREQ=YEARLY;BYMONTH=1
TZNAME:SAST
TZOFFSETFROM:+0200
TZOFFSETTO:+0200
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-adass-xxxi-2021-8XXCYS@pretalx.adass2021.ac.za
DTSTART;TZID=Africa/Johannesburg:20211028T120000
DTEND;TZID=Africa/Johannesburg:20211028T121500
DESCRIPTION:Modern astronomical telescopes observe millions of transients e
very night. Light curves of these objects help to deduce their physical an
d phenomenological properties. A light curve is a flux time series measure
d in one or several photometric passbands. Each passband might have a diff
erent number of observations and various time gaps between them\, thus com
plicating further analysis. To overcome these difficulties\, we propose ne
w methods for light curve approximation based on neural network models. \n
\nIn this study\, we use The Photometric LSST Astronomical Time Series Cla
ssification Challenge (PLAsTiCC) dataset. We apply shallow neural networks
\, bayesian neural networks\, and normalizing flows to approximate observa
tions for a single object and facilitate analysis of the light curves. We
show that the approximation quality of the proposed methods significantly
outperform the existing approaches based on gaussian processes. We also pe
rform two physical analyses: supernovae type Ia classification and intensi
ty peak estimation. For both these problems\, convolutional neural network
s are trained on approximated light curves. The results show that the prop
osed methods help to improve the quality of supernovae type identification
and increase the accuracy of the intensity peak estimation compared with
the gaussian processes model.
DTSTAMP:20221209T215813Z
LOCATION:Grand Ballroom
SUMMARY:Astronomical Data Approximation Based on Neural Network Models - Sa
morodova Ekaterina\, Konstantin Malanchev
URL:https://pretalx.adass2021.ac.za/adass-xxxi-2021/talk/8XXCYS/
END:VEVENT
END:VCALENDAR