Artificial intelligence finds 56 new gravitational lens candidates

October 23, 2017, Netherlands Research School for Astronomy
candidates. This picture shows a sample of the handmade photos of gravitational lenses that the astronomers used to train their neural network. Credit: Enrico Petrillo, University of Groningen

A group of astronomers from the universities of Groningen, Naples and Bonn has developed a method that finds gravitational lenses in enormous piles of observations. The method is based on the same artificial intelligence algorithm that Google, Facebook and Tesla have been using in the last years. The researchers published their method and 56 new gravitational lens candidates in the November issue of Monthly Notices of the Royal Astronomical Society.

When a galaxy is hidden behind another galaxy, we can sometimes see the hidden one around the front system. This phenomenon is called a gravitational lens, because it emerges from Einstein's general relativity theory which says that mass can bend light. Astronomers search for because they help in the research of dark matter.

The hunt for gravitational lenses is painstaking. Astronomers have to sort thousands of images. They are assisted by enthusiastic volunteers around the world. So far, the search was more or less in line with the availability of new images. But thanks to new observations with special telescopes that reflect large sections of the sky, millions of images are added. Humans cannot keep up with that pace.

Google, Facebook, Tesla

To tackle the growing amount of images, the astronomers have used so-called 'convolutional neural networks'. Google employed such neural networks to win a match of Go against the world champion. Facebook uses them to recognize what is in the images of your timeline. And Tesla has been developing self-driving cars thanks to .

The astronomers trained the neural network using millions of homemade images of gravitational lenses. Then they confronted the network with millions of images from a small patch of the sky. That patch had a surface area of 255 square degrees. That's just over half a percent of the sky.

Artificial intelligence finds 56 new gravitational lens candidates
With the help of artificial intelligence, astronomers discovered 56 new gravity lens candidates. At this picture are three of those candidates. Credit: Carlo Enrico Petrillo, University of Groningen

Gravitational lens candidates

Initially, the neural network found 761 gravitational lens candidates. After a visual inspection by the astronomers the sample was downsized to 56. The 56 new lenses still need to be confirmed by telescopes as the Hubble space telescope.

In addition, the neural network rediscovered two known lenses. Unfortunately, it did not see a third known lens. That is a small and the neural network was not trained for that size yet.

In the future, the researchers want to train their neural network even better so that it notices smaller lenses and rejects false ones. The final goal is to completely remove any visual inspection.

Kilo-Degree Survey

Carlo Enrico Petrillo (University of Groningen, The Netherlands), first author of the scientific publication: "This is the first time a has been used to find peculiar objects in an astronomical survey. I think it will become the norm since future astronomical surveys will produce an enormous quantity of data which will be necessary to inspect. We don't have enough astronomers to cope with this."

The data that the neuronal network processed, came from the Kilo-Degree Survey. The project uses the VLT Survey Telescope of the European Southern Observatory (ESO) on Mount Paranal (Chile). The accompanying panoramic camera, OmegaCAM, was developed under Dutch leadership.

Explore further: Discovery of a rare quadruple gravitational lens candidate with Pan-STARRS

More information: C. E. Petrillo et al. Finding strong gravitational lenses in the Kilo Degree Survey with Convolutional Neural Networks, Monthly Notices of the Royal Astronomical Society (2017). DOI: 10.1093/mnras/stx2052

Related Stories

Most distant gravitational lens helps weigh galaxies

October 17, 2013

An international team of astronomers has found the most distant gravitational lens yet—a galaxy that, as predicted by Albert Einstein's general theory of relativity, deflects and intensifies the light of an even more distant ...

Recommended for you

Where to search for signs of life on Titan

July 20, 2018

New findings, published in the journal Astrobiology, suggest that large craters are the prime locations in which to find the building blocks of life on Saturn's largest moon, Titan.

Did a rogue star change the makeup of our solar system?

July 20, 2018

A team of researchers from the Max-Planck Institute and Queen's University has used new information to test a theory that suggests a rogue star passed close enough to our solar system millions of years ago to change its configuration. ...

Traveling to the sun: Why won't Parker Solar Probe melt?

July 19, 2018

This summer, NASA's Parker Solar Probe will launch to travel closer to the Sun, deeper into the solar atmosphere, than any mission before it. If Earth was at one end of a yard-stick and the Sun on the other, Parker Solar ...

12 comments

Adjust slider to filter visible comments by rank

Display comments: newest first

Hyperfuzzy
1 / 5 (14) Oct 23, 2017
Are there any real scientist? To modify the light a media is required. Fields don't affect frields. Gravity is the field due to all the charges that exist from their centers to infinity. Any "mass" or group of charges are attracted to other groups but these create no lensing. A necessary and sufficient conditions for lensing is a lens like media. Juz say'n This a confusion of how gravity is defined from known information, not some erroneous application of a scalar.

My hobby is theoretical physics, every since I was thirteen. I remember the speculation upon Steady State and The Big Bang; however, observations and reality is much simpler!
RogueParticle
4.7 / 5 (14) Oct 24, 2017
@HF - how nice that you have a hobby in theoretical physics. I have a friend whose hobby is building FTL spacecraft: he's having trouble right now deciding just which particular shade of pink he should paint them. Perhaps you could help him with your in-depth knowledge of physics and stuff? Juz say'n...
Repurplecirculation
5 / 5 (2) Oct 24, 2017
I presume they incorporated light spectra, not just shapes, for the training data. A true gravitational lens will not alter the wavelengths of light (it's not like a prism - there can be no diffraction). Any image where the light is of a different frequency when "lensed" vs when "unlensed" (the source behind it), is not due to gravity.
antialias_physorg
5 / 5 (8) Oct 24, 2017
because it emerges from Einstein's general relativity theory which says that mass can bend light

I'm always a bit queasy when I see it put like this. Light doesn't bend. Light travels in a straight line (it's more or less how 'straight' is *defined*: The path of least time taken by light from point A to point B is the straight path).
It's spacetime that's warped.

The bending is only how it appears to us because our brains are incapable of considering the intervening space as being anything but euclidian. Even in such simple circumstances as a straw that sticks halfway out of the water and appears 'bent' we are incapable of actually seeing the true form - even though we, intellectually, know for a fact that the straw is straight.
Hyperfuzzy
1 / 5 (4) Oct 24, 2017
@HF - how nice that you have a hobby in theoretical physics. I have a friend whose hobby is building FTL spacecraft: he's having trouble right now deciding just which particular shade of pink he should paint them. Perhaps you could help him with your in-depth knowledge of physics and stuff? Juz say'n...

Lime Green Florescent with paint that absorbs the vast amount of energy produced while moving through the fields and also stores the energy from any charged particles. For molecular collisions, there's built in damper field by applying a small predefined signal unto the baseplate. The paint also provides digital images from every molecule of the paint. The precoat applies connectivity with a fuzzy algorithm the coating thinks for it self and is also the ships computer, auxiliaries and fail safe by design. This coating is also the energy source for everything including the impulse engines and the hyperdrive. The ship responds to the hyper-drive by placing a field throughout
Hyperfuzzy
1 / 5 (3) Oct 24, 2017
to simulate any gravity field. Of course there's more you might need to learn before we can communicate.
TrollBane
5 / 5 (8) Oct 24, 2017
The Hyperfuzzy space flight program, achieving head-posterior separation at T minus never.
Hyperfuzzy
1 / 5 (2) Oct 24, 2017
The Hyperfuzzy space flight program, achieving head-posterior separation at T minus never.

You're correct, present science interpret these distortion with "nonsense speak". This light "may" be light that has passed through our galaxy or neighbors. Calculate the vector space without the nonsense. And please, be relevant to the science of light! Only created by charge motion. The field ...
PowerMax
not rated yet Oct 25, 2017
@HF - few ideas but well confused
Da Schneib
5 / 5 (2) Oct 25, 2017
Looks like they found another way to get this job done: https://phys.org/...ion.html

Wonder which one's better?
Hyperfuzzy
1 / 5 (2) Oct 25, 2017
Looks like they found another way to get this job done: https://phys.org/...ion.html

Wonder which one's better?

There is no logic when we dismiss logic.
TrollBane
not rated yet Oct 25, 2017
Let's get a scope out to our own sun's focal point.

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.