MENU

SETI Deploys AI To Help Search For Alien Life

Close

SETI is supercharging the search for alien life by using artificial intelligence to analyze dozens of data collected from a galaxy far, far away.

A Big Mystery

The researchers from Breakthrough Listen, a project led by the University of California Berkely, used a new neural network to discover 72 fast radio bursts or FRB from a mysteriously noisy galaxy about three billion light-years away from Earth.

"This work is exciting not just because it helps us understand the dynamic behavior of fast radio bursts in more detail, but also because of the promise it shows for using machine learning to detect signals missed by classical algorithms," stated  Andrew Siemion, Berkeley SETI Research Center director.

FRB are pulses of radio emissions, which last about a millisecond, have remained a mystery to this day. Plenty of theories try to explain their origin including strong magnetic fields in a dense plasma. Some believe that these FRB might be coming from a technology developed by an advanced alien civilization.

FRB 121102, the subject of the study published in The Astrophysical Journal, is one stellar object that has drawn the attention of scientists. Unlike other FRBs, FRB 121102 is not a one-off event. It is emitted by its mysterious source repeatedly from billions of light years away.

To analyze the signals, researchers recorded data over a five-hour period with the help of the Green Bank Telescope in West Virginia in August 2017. The session yielded 400 terabytes of transmission data.

An initial analysis using the standard computer algorithm was able to identify 21 radio bursts within one hour of the observation.

Using AI

Researchers wanted to reanalyze the data using the same techniques that internet companies use to optimize search engine results and classify images. Gerry Zhang, a graduate student at UC Berkeley, developed the "convolutional neural network system" to scour the massive amount of data and find radio bursts from FBR 121102.

The new algorithm was able to find an additional 72 bursts from the recorded transmission data, bringing the total of bursts recorded from FBR 121102 to around 300 since its discovery in 2012.

The researchers did not find evidence suggesting that the emission was from an artificial origin from a planet far away. They detected no pattern from the bursts.

However, the new method developed by Zhang and his team changes the way scientists gather fast radio bursts that might help uncover the mystery of their origin in the future.

"We hope our success may inspire other serious endeavors in applying machine learning to radio astronomy," Zhang stated.

ⓒ 2018 TECHTIMES.com All rights reserved. Do not reproduce without permission.
Real Time Analytics