Friday, March 29, 2024
HomeSECURITYCanadian students have developed an ML model to detect signals from space

Canadian students have developed an ML model to detect signals from space

-


Canadian students have developed an ML model to detect signals from space

The technology will help scientists find previously undiscovered cosmic signals.

Scientists have developed a machine learning method that can help researchers filter out interference and more effectively detect unusual radio signals from space, helping them search for extraterrestrial intelligence.

The Search for Extraterrestrial Intelligence (SETI) programs have used radio telescopes for decades to detect cosmic signals. However, this search has become more complicated due to the development of technologies that can generate false positives that require a lot of time to filter from large datasets.

SETI scientists modeled many signals, implemented them into real observations, and trained the ML system to classify these simulations. The autoencoder part is trained both on real observations and on simulations while recreating the original input data, which helps specialists to highlight the salient features of the input image. Together, this helps create an efficient anomaly detection algorithm.

AT research under the guidance of Peter Ma, a 3rd year student in the Faculty of Physics and Mathematics at the University of Toronto, observations of 820 stars were used in the form of 115 million pieces of data. The ML models developed by the team using the TensorFlow machine learning library and the Python Keras library identified about 3 million signals of interest. The set was then reduced to 20,515 signals, which is more than 100 times less than in the previous analysis of the same data set.

Scientists have identified 8 previously undetected signals, although these signals were not displayed in subsequent observations. The authors suggest that their method can be applied to other large datasets to speed up SETI and similar data-driven surveys.

Other developed ML applications at SETI include:

  • a common signal classifier for observations obtained with the Allen telescope array and with a spherical radio telescope with a 500-meter aperture;
  • RF interference identifiers based on convolutional neural networks;
  • anomaly detection algorithms.

“SETI aims to find evidence of intelligent life in space using “technosignatures” created by technology scientists. The discovery of an unambiguous technosignature will demonstrate the existence of extraterrestrial intelligence (ETI), so this is of great interest to both scientists and the general public,” the researchers said.



Source link

www.securitylab.ru

LEAVE A REPLY

Please enter your comment!
Please enter your name here

Most Popular