Seminarium instytutu

Zapraszamy na seminarium w poniedziałek 9 grudnia w godz. 12:15 – 13:15 w sali 3/40, bud. 34.

Referent: Prof. Amitava Datta, The University of Western Australia

Tytuł : Applications of deep learning in Astrophysics

Abstract: Astronomical data has exploded in the last few decades, generated both by earth-based and space-based telescopes. Moreover, large-scale surveys are used for collecting as much data as possible within the short life times of the missions. Traditional data analysis tools are inadequate for analysing such data. We will discuss in this talk how deep learning (and machine learning in general) is revolutionising analysis of astrophysical data. We will give examples from several different areas of astrophysics, including localisation of gravitational wave sources, search for exoplanets, estimating galactic masses from rotations of globular clusters, photometric redshift and detecting transients like Type 1A and Type II supernovae. In each case, I will explain the background and discuss the deep learning architectures and training methodologies. We will see that simple neural network models like Multilayer Perceptron (MLP), Convolutional Neural Networks (CNN) and Recurrent Neural Networks (RNN) achieve very high accuracy and low false positive rates in these tasks.