Search for Extraterrestrial Intelligence, or SETI, is the newest research organization to deploy AI to help manage data sets and surprisingly, it got instant results. The AI algorithms managed to discover seventy-two new “fast radio bursts” from a mysteriously noisy galaxy 3 billion away. These radio bursts were discovered in previously analyzed data by using a custom machine learning model.
The researchers note that this is not Morse code or encrypted instructions to build contact but these fast radio bursts or FRBs are poorly understood. At the very least, these FRBs could represent some unobserved cosmic phenomenon. FRB 121102 is the only stellar object known to give off the signals regularly and is the target of continued observation.
The data comes from the Green Bank Telescope in West Virginia, which pointed toward this source of fast and bright bursts for five hours in August of 2017. TechCrunch reports that the five-hour session yielded 400 terabytes of transmitted data. The report adds that initial ‘standard’ algorithms identified 21 FRBs that happened during an hour long observation.
The convolutional neural network system created by Gerry Zhang, a graduate student at UC Berkeley and part of the Breakthrough Listen project, was able to pick out 72 more FRBs in the same period. For any neural network to be successful, there is need for initial data and with the data acquired by these tests, the system has a tool capable of collecting data.
The paper that details the discoveries will be co-authored by Vishal Gajjar, Postdoctoral Researcher at UC Berkeley and will be published in the Astrophysical Journal. The new data also reveals that the signals are not being received in any kind of pattern that can be determined. While it may sound discouraging, it will help researchers better understand these fast bursts of sound.
Watch: Apple iPhone XS, XS Max, and XR First Look
SETI and Breakthrough plan to keep their antennas tuned to FRB 121102 and will collect data that will help in future science. We will know more about the work once the paper is published in the Astrophysical Journal.