New neural network catalogs 301 additional exoplanets



(Photo: Getty Images)



The art of the neural network
(Photo: Getty Images)



exoplanet art
(Photo: Getty Images)

A new neural network called ExoMiner managed to add a total of 301 new exoplanets, bringing our total to 4,569.

exoplanet art

(Photo: Getty Images)

According to Phys.org, the aforementioned neural network used NASA’s incredibly powerful Pleiades supercomputer to confirm the existence of all 301 exoplanets.

All of these planets were originally discovered using data from the Kepler archive. But until the use of ExoMiner, these were only classified as potential exoplanets and could not be confirmed.

Using the power of NASA’s Pleiades, the ExoMiner neural network was able to verify the planetary status of new discoveries. Then it was determined that these exoplanets, while interesting on their own, do not resemble Earth or are not located within the habitable zones of their parent stars.

How accurate is the neural network?

The ExoMiner neural network is super accurate. According to Cornell University, the network has achieved an accuracy of 93.6% (the maximum is 99%), which makes it much more accurate and reliable than current classifiers that do not use machine learning technology. .

The art of the neural network

(Photo: Getty Images)

All thanks to the power of NASA’s Pléiades supercomputer, which is one of the most advanced in the world. Indeed, a neural network would be nothing without hardware acceleration.

According to NASA’s official specifications, Pleiades is capable of producing 7.09 petaflops of peak cluster performance. It has a total of 241,324 CPU cores, 927TB of total memory, and 614,400 CUDA GPU cores, well beyond the specs of your regular high-end gaming PC.

There is such confidence in the capabilities of the ExoMiner neural network. Project leader Hamed Valizadegan is confident enough to say that when the neural network says it’s an exoplanet, it’s best to believe it is, according to the original Phys.org report.

Read also: The new class of exoplanets called “Hycean Worlds” is very promising for habitability

ExoMiner and planet hunting

Astronomers previously had another way to search for exoplanets: the transit method. This technique, according to Universe Today, allows scientists to spot extrasolar planets by measuring the light emitted by their parent star.

When a planet passes in front of its star, it is called a transit. A transit will always cause a slight drop in its luminosity. Astronomers will then try to determine if the dip occurs at regular intervals.

 Exoplanet in front of the sun

(Photo: Getty Images)

If so, then they might be sure that the transiting celestial body is, in fact, a planet. That’s because anything can transit a star – an asteroid, a moon, or whatever. Only a real planet will transit at regular intervals.

The only problem with this method is that there is no real way to verify the planetary status of any celestial bodies detected. This is where the ExoMiner neural network comes in.

Look forward

For now, the neural network has been trained using current data from the Kepler mission. Armed with this information, he will now use everything in his power to further verify any new exoplanets he encounters.

Additionally, the researchers say there is still room for improvement in the machine learning technique, which could lead to even more exciting discoveries in the future.

Related article: AI-Based Neural Network Brings Photorealistic Edits to Elon Musk, Mark Zuckerberg Photos

This article belongs to Tech Times

Written by RJ Pierce

ⓒ 2021 TECHTIMES.com All rights reserved. Do not reproduce without permission.