Abstract: Breakthrough Listen is the largest effort to date to search for techno-signatures from extraterrestrial civilizations. We use extensive computing power to search at high frequency and time resolution for transients events in petabytes of observational data from the Green Bank Telescope, Parkes Telescope, LOFAR, and soon MeerKAT. Given the diverse manifestations of transient signals in an environment of increasing anthropogenic RFI, machine learning-based models are proving essential in filtering and detecting anomalous signals. We are actively using classic wide-feature models and deep neural network models to detect and classify astrophysical signals such as pulsars and FRBs. We are building large labelled datasets from a diverse sample of observations in order to facilitate ML research groups in developing new algorithms which we can incorporate into our search pipelines.
Bio: Griffin is a post-doctoral researcher in the physics department at the University of Oxford and a visiting scholar at the University of California at Berkeley. He is the project scientist for Breakthrough Listen on MeerKAT. He was previously an SKA Research Fellow at Rhodes University and SARAO in South Africa where he worked on interferometric calibration and imaging. His D.Phil thesis (Oxon 2013) was on FPGA-based correlator design and low-frequency aperture synthesis.