Adam Zadrożny, an Assistant Professor at National Centre for Nuclear Research (NCBJ), Poland. He holds Ph.D. in Physics (2015) from NCBJ. His thesis was focused on multi-messenger astronomy. In 2017-2018 he made his Post-Doc at Center for Gravitational Wave Astronomy at The University of Texas Rio Grande Valley, USA working on AI for astrophysical data. He hold M.Sc. degree in theoretical physics from University of Warsaw (2008). He has done internships or projects for Interdisciplinary Center for Mathematical Modeling of University of Warsaw, Prokom Software S.A. and Facebook Inc..
Using Convolutional Neural Networks along image subtraction for detecting weak optical transients
In the poster I would like to present a method for enchasing image subtraction with CNN for detecting optical transients. The method has been tested on Transient Optical Robot Observatory of the South (TOROS) optical data connected with search of optical counterparts to gravitational wave events during LIGO O2 run. In test runs method proved to achieve more than 99% accuracy, much higher than algorithms based on random forest methods. The method will be included into standard transient search pipeline for TOROS network in upcoming year.