Can machine learning be used in the hunt for life beyond Earth? To test this theorem, scientists have applied a deep learning technique to a previously studied dataset of nearby stars. This exercise has uncovered eight previously unidentified signals of interest.
The hunt for extraterrestrial life, as measured by life on another planet developing some form of technology that might eb detectable ,has been hampered by limited technology and algorithms developed decades ago. These established approaches cannot reliably process modern petabyte-scale datasets.
As an alternative, a deep learning technique holds greater promise. To examine this, the application was commanded to search through 150 TB of data of 820 nearby stars. This was using a dataset that had previously been searched through by classical techniques but labelled as devoid of interesting signals.
The search for extraterrestrial intelligence (SETI) looks for evidence of extraterrestrial intelligence originating beyond Earth by trying to detect technosignatures (any measurable property or effect that provides scientific evidence of past or present technology) that alien civilizations could have developed. The most common technique is to search for radio signals. SETI experiments began in 1960 with Frank Drake’s Project Ozma at the Greenbank Observatory.
This study re-examined data taken with the Green Bank Telescope in West Virginia as part of a ‘Breakthrough Listen’ campaign that initially indicated no targets of interest. The goal was to apply new deep learning techniques to a classical search algorithm to yield faster, more accurate results.
After running the new algorithm and manually re-examining the data to confirm the results, newly detected signals had several important characteristics.
First of all, the signals were of a narrow spectral width, on the order of just a few Hz. Signals caused by natural phenomena tend to be broadband.
One of the researchers, Cherry Ng, said: “These results dramatically illustrate the power of applying modern machine learning and computer vision methods to data challenges in astronomy, resulting in both new detections and higher performance. Application of these techniques at scale will be transformational for radio technosignature science.”
The new computational tools show greater capacity for processing and analysing that data quickly to identify anomalies that could be evidence of extraterrestrial intelligence.
The research appears in the journal National Astronomy, titled “A deep-learning search for technosignatures from 820 nearby stars.”
