5th Seminar on Gravitational Waves and Numerical Relativity


  • Date: Dec. 7 (Fri.), 2018, 2 pm - 5 pm
  • Venue: Room 102, Lee Wonchul Hall, KASI
  • Title: Applicaton of Machine Learning to Detecting Gravitational-Wave Signal
  • Speaker: Kyungmin Kim (CUHK)
  • Abstract: Conventional searches for gravitational waves radiated from mergers of compact binary systems produce candidate event data based on matched filtering. The candidate data contains various information for example signal-to-noise ratio, physical parameters of template waveforms of matched filtering, results of signal consistency tests, and so on. Thus, we can regard searching gravitational-wave signals from binary mergers as a multivariate problem and this assumption seamlessly leads to consider machine learning as an alternative method to the searches. In this seminar, we present two preliminary results of two different works on applying machine learning to low-latency gravitational-wave search and to identifying lensed gravitational waves. For the low-latency search, we show the result of a feasibility study on using the output of machine learning as a new raking method for the production of candidate events. While, for the identification of lensed gravitational waves, we present how to estimate physical parameters of the lensed waves from the spectrogram of gravitational-wave signal.
  • Slide File 1 Slide File 2 Slide File 3


seminar02 seminar02 seminar02 seminar02