Classifying Periodic Astrophysical Phenomena from non-survey optimized variable-cadence observational data
作者：
Highlights：
• Instruments on the Liverpool Telescope have produced inexpensive variable-cadence wide-field data.
• Variable-cadence introduces artifacts into the data which can produce false positives.
• Light curve features quantify variability class but current ones are susceptible to artifacts.
• Using representation learning, new features can be learnt focused on the real astrophysical signals.
• Trained classifiers indicate the new features are superior to previous approaches on this data.
摘要
•Instruments on the Liverpool Telescope have produced inexpensive variable-cadence wide-field data.•Variable-cadence introduces artifacts into the data which can produce false positives.•Light curve features quantify variability class but current ones are susceptible to artifacts.•Using representation learning, new features can be learnt focused on the real astrophysical signals.•Trained classifiers indicate the new features are superior to previous approaches on this data.
论文关键词：Astronomical time-series,Light curve analysis,Period analysis,Variable stars,Binary stars,Random forest classification
论文评审过程：Received 1 March 2018, Revised 10 March 2019, Accepted 16 April 2019, Available online 17 April 2019, Version of Record 30 April 2019.
论文官网地址：https://doi.org/10.1016/j.eswa.2019.04.035
