Machine learning is used in many fields, including in the projects SETI for search for extraterrestrial civilizations. Researchers at the University of California at Berkeley used it in the framework of the project a Breakthrough Listen to study signals from a mysterious source located at a distance of 3 billion light years from Earth.
Signals represent radiospace a duration of several milliseconds. Their origin is not exactly known, but it is assumed that they are emitted by distant galaxies. One theory says that outbreaks occur in the explosion of a strongly magnetized neutron stars near a supermassive black hole. Also, it is believed that the signals of distant civilizations.
The neural network Breakthrough Listen studied 700 TB data collected by the radio telescope green Bank on 26 August 2017. It has identified 72 flash from galaxy FRB 121102, and the former methods were found only 21 signal.
The basis of the algorithm was a Breakthrough Listen convolutional neural network, typically used for recognition of objects in images. In the framework of the project she was looking for outbreaks of noise.
This project is only the beginning of the use of neural networks for detection of signals. We hope that our success will inspire other researchers to use machine learning in astronomy. — announced project participant Jerry Zhang (Gerry Zhang).
Examining the frequency of outbreaks, scientists have found that signals FRB 121102 is no logical order. Even if this is not the message from other civilizations, further study will help to expand our understanding of the universe.
A theory of the origin of mysterious signals you adhere to? An opinion can be expressed in our telegram chat.
The neural network has detected signals from a mysterious source
Ramis Ganiev