Symposium S11  30 June - 1 July 2022

Machine Learning: a giant leap towards space discovery in the era of peta and exabyte scale surveys

News: You find the program of the symposium at the following links: Day 1, Day 2

Aims and scope

Astronomy and cosmology are at the forefront of Big Data science, with exponentially growing data volumes, rates and complexity. Data from current and future generation 'Big Data' missions (e.g. Gaia, eROSITA, Euclid, JWST, Roman, Athena, Rubin-LSST) and telescopes (SKA, ELT, 4MOST) will reach the peta and exabyte regimes and become the new normal in future astrophysical research. These datasets will require game-changing statistics and provide high-quality data that is essential for tackling key open questions such as the energy and mass content of the Universe or the major physical quantities that drive galaxy formation. However, the extraordinary volume of these datasets will present novel challenges as data volumes at these scales have never been encountered by the scientific community before. Such challenges include identifying different types of astronomical objects over short timescales, large-scale morphological analysis of galaxies, time-domain variability predictions, visualization techniques and many more. In anticipation of Big Data surveys, the community has increasingly invested in developing advanced machine learning (ML) and artificial intelligence (AI) based data-analysis techniques, which can provide the level of accuracy and automation required for the efficient exploitation of extremely large datasets. Such methods are rapidly becoming the default choice for many astronomers. In fact, AI allows us to analyse huge amount of data for the most disparate set of scientific aims, from cosmology, to stellar physics, extragalactic astronomy, and planetary science. Furthermore, exascale computing will provide the capability to tackle challenges in scientific discovery at levels of complexity and performance that previously were out of reach, thus enabling the full exploitation of multi-messenger Astrophysics and numerical simulations for astrophysics and cosmology. The aim of this symposium is to bring together astronomers and data scientists working to apply AI or ML techniques to astronomical questions, with a particular focus on cosmology and galaxy evolution. Given the imminent arrival of extremely large surveys, this meeting will be timely and enable researchers to build collaborations that are likely to benefit the astronomical community for many years to come. The key points we want to discuss are the following:

  • What is the current state-of-the-art for ML applications in existing datasets and what have we learned so far?
  • What new developments in computer science, astronomy or other scientific contexts can we learn from? This includes (but not limited to) source and anomaly detection, deblending techniques and morphological analysis.
  • For which applications is a data driven approach useful, in particular for parameter estimation in galaxies (e.g. photometric redshift, stellar mass and star formation rate from photometry and spectra, and dark matter profile from dynamical data, gravitational lensing, etc.)?
  • In most of the real world applications, ML algorithms are left free to explore the data parameter space, searching for hidden correlations. How does our matured knowledge on data and related astrophysics impact performance?
  • We can discover new rare sources (e.g. gravitational lenses or new planets) by determining their physical parameters with AI, but can such approaches allow us to better understand the physical processes driving planet, star and galaxy evolution and the energy and mass content of the Universe?
  • With Euclid and Rubin almost ready for operation, and other instruments soon to follow, we will soon have a much more detailed picture of dark matter, dark energy and galaxy evolution across cosmic time. What are the tools and future plans can we implement to exploit these these big datasets to their fullest potential?

Programme

This Symposium is organized in 6 blocks of 1.5 hours each.

Here it is a broad list of topics that will be covered during the symposium:

  • Exploitation of datasets in wide-field surveys (e.g. Gaia, eROSITA, Euclid, JWST, Roman, Athena, Rubin-LSST, etc.)
  • Galaxy properties (morphology, stellar population parameters, kinematics, etc)
  • Photometric redshifts
  • Gravitational lensing in galaxies and clusters
  • Time series analysis
  • Stellar physics
  • Solar system and exoplanets search
  • Cosmological simulations with machine learning
  • Applications of machine learning techniques to cosmology
  • Applications to radio-astronomy
  • Multi-Messenger astronomy
  • General Relativity and gravitational waves
  • Connection with citizen science
  • New methodologies for astronomical and cosmological applications

Invited speakers

  • Bunte Kerstin (University of Groningen, The Netherlands)
  • Domínguez-Sánchez Helena (IEEC-CSIC, Spain)
  • Grillo Claudio (Università di Milano, Italy)
  • Navarro-Villaescusa Francisco (Princeton University, US)
  • Ntampaka Michelle (Space Telescope Science Institute, US)

Scientific organisers

  • Brescia M. (chair, INAF-OAC, Italy)
  • Chamba N. (chair, Stockholm University, Sweden)
  • Lazar I. (chair, University of Hertfordshire, UK)
  • Martin G. (chair, KASI, Korea / Steward Observatory, USA)
  • Tortora C. (chair, INAF-OAC, Italy)
  • Borgani S. (INAF-OATS, Italy)
  • Huertas-Company M. (Observatory of Paris, France)
  • Jaodand A. (Caltech, USA)
  • Kaviraj S. (University of Hertfordshire, UK)
  • Mei S. (Observatory of Paris, France)
  • Salvato M. (MPE, Germany)
  • Sarmiento R. (IAC, Spain)
  • Scaife A. (University of Manchester, UK)
  • Wenskovitch J. (Virginia Tech, USA)

Contact

  • Massimo Brescia: massimo.brescia @ inaf.it
  • Nushkia Chamba: nushkia.chamba @ astro.su.se
  • Ilin Lazar: i.lazar @ herts.ac.uk
  • Garreth William Martin: garrethmartin @ arizona.edu
  • Crescenzo Tortora: crescenzo.tortora @ inaf.it

    • Updated on Sat Jun 11 12:04:19 CEST 2022
European Astronomical Society