Universiteit Leiden

nl en

Artificial brain helps Gaia satellite catch speeding stars

With the help of software that mimics a human brain, ESA’s Gaia satellite spotted six stars zipping at high speed from the centre of our Galaxy to its outskirts. This could provide key information about some of the most obscure regions of the Milky Way.

Speed by supermassive black hole

A new class of high-speed stars was discovered just over a decade ago. Swooping through the Galaxy at several hundred of km/s, they are the result of past interactions with the supermassive black hole that sits at the centre of the Milky Way and, with a mass of four million Suns, governs the orbits of stars in its vicinity.

Crucial information

‘These hypervelocity stars are extremely important to study the overall structure of our Milky Way,’ says Elena Maria Rossi from Leiden Observatory, who presented Gaia’s discovery of six new such stars at the European Week of Astronomy and Space Science in Prague on 26 June. ‘These are stars that have travelled great distances through the Galaxy but can be traced back to its core – an area so dense and obscured by interstellar gas and dust that it is normally very difficult to observe – so they yield crucial information about the gravitational field of the Milky Way from the centre to its outskirts.’ According to Rossi, these stars can be used to measure the dark matter, which makes up most of the Galaxy content.

Learning from experiences

The billion-star recording being performed by Gaia offers a unique opportunity, so Elena and her collaborators started wondering how to use such a vast dataset to optimise the search for fast-moving stars. After testing various methods, they turned to software through which the computer learns from previous experience.

Mimicking the brain

‘In the end, we chose to use an artificial neural network, which is software designed to mimic how our brain works,’ explains Tommaso Marchetti, PhD student at Leiden Observatory and lead author of the paper describing the results published in Monthly Notices of the Royal Astronomical Society. ‘After proper training, it can learn how to recognise certain objects or patterns in a huge dataset. In our case, we taught it to spot hypervelocity stars in a stellar catalogue like the one compiled with Gaia.’


'This result showcases the great potential of Gaia opening up new avenues to investigate the structure and dynamics of our Galaxy,' says Anthony Brown , a co-author on the study and chair of the Gaia Data Processing and Analysis Consortium. The scientists are looking forward to using data from the next Gaia release, which is planned for April 2018 and will include distances and motions on the sky for over a billion stars, as well as velocities for a subset.

Full ESA press release 

Photocredits: ESA, CC BY-SA 3.0 IGO 

This research has been funded by the Netherlands Organisation for Scientific Research (Nederlandse Organisatie voor Wetenschappelijk Onderzoek, NWO).