AI Could Send Astronomy Into Hyperdrive Mode

Read Time:3 Minute, 2 Second
THE MOST IMPORTANT OVERVIEW

Due to the huge amounts of data generated by the exploration of the universe, artificial intelligence is undeniably of great importance for the future of astronomy. It will be a cornerstone in deepening our understanding of space.

The universe, the vaunted ultimate frontier, is the last gigantic ocean of exploration that lies ahead, and perhaps artificial intelligence (AI) will provide us with bigger paddles.

The Vera C. Rubin Observatory in Chile is a significant step on this path. Over the next decade, the observatory will generate 0.5 exabytes of data .

How much is this? About 50,000 times more than all the books in the Library of Congress in Washington combined. So a lot.

The observatory has 20 telescopes with mirrors that are more than 20 feet in diameter, allowing them to capture the smallest reflections of light from light-years away.

The data is important, but only as long as astronomers can process it and interpret it meaningfully.

This is where artificial intelligence comes into play because it can search through large structured data sets and identify patterns or anomalies much faster than humans.

Let there be light

The Rubin Observatory’s mission is to ” build a well-understood system that will produce an unprecedented astronomical data set for the study of the deep and dynamic universe, make the data available to a diverse community of scientists, and the public to explore the universe.” to move with us . ”

Machine learning and neural networks are increasingly being used to sift through the data. This has already happened, e.g. B. with the dictionary learning-based PRIMO algorithm , which uses high-quality black hole simulations as a data set.

It is difficult to detect galaxies in an astronomical image: almost 99% of the light captured comes from other sources or from radiation in the background, and only 1% may be the light from distant galaxies.

Doing this manually is either impossible or very time-consuming.

Galaxies as far as the eye can see (and beyond)

AI algorithms that use neural networks with many interconnected nodes can be used to identify galaxies, with studies showing up to 98% detection accuracy .

This was the result at the Subaru Telescope, where astronomers applied AI to ultra-wide field-of-view images of the distant universe and achieved very high precision in finding and classifying spiral galaxies in these images.

A research group composed primarily of astronomers from the National Astronomical Observatory of Japan (NAOJ) applied deep learning to classify galaxies in a large image dataset.

Thanks to the system’s high sensitivity, up to 560,000 galaxies were discovered in the images. The AI ​​allowed the team to carry out the processing without human intervention.

Thanks to neural networks, astronomers can now identify distant civilizations on planets – if they exist.

Groups such as the Search for ExtraTerrestrial Intelligence (SETI) and instruments such as radio telescopes are used in the search for extraterrestrial intelligence.

More recently, they have been supported by AI neural networks that used 150,000 PCs and 1.8 million citizen scientists to decode radio signals .

While there is no confirmation of life out there yet, the ability to scan and process huge data sets will significantly speed up research while filtering out false positive signals .

Conclusion

The truth may be out there. But we will most likely find them first in the vast data sets that can be collected with ever larger and more powerful telescopes on Earth and in space.

The tantalizing combination of data and AI could send the search for everything – planets, galaxies, black holes, the galactic empire – into overdrive.

Happy
Happy
0 %
Sad
Sad
0 %
Excited
Excited
0 %
Sleepy
Sleepy
0 %
Angry
Angry
0 %
Surprise
Surprise
0 %