• Time: Thursday, October 5th, 2017 at 05:30pm
  • Place: WEH 8220

Abstract

The main challenge of Cosmology in the 21st century is to understand the nature of Dark Matter and Dark Energy, which combined seem to account for a staggering 95% of our Universe despite remaining complete mysteries. To shed some much needed light on these questions, Cosmologists rely on a variety of probes such as measuring the Cosmic Microwave Background (relic radiation from the early Universe) or the gravitational lensing effect (deflection of light rays by massive structures). However, extracting cosmological information from the raw data collected by current or next generation surveys involves a large number of statistical and signal processing challenges. In this talk I will first present applications of sparse signal representation (e.g. wavelets) to address some of these challenges, in particular to solve large scale inverse problems (Compressed Sensing theory). I will also present some exciting applications of deep learning, including very recent developments based on graph theory.


Pizzas will be served.