Machine Learning on Breakthrough Listen Data from the Green Bank Telescope: Anomaly Detection for Technosignatures with No Confirmed Detections
To the point
Machine learning was applied to Green Bank Telescope data to search for technosignatures, filtering out Earth-based interference and flagging signals of interest, but follow‑up failed to re-detect them so they are not bona fide candidates, and the approach remains promising for spotting anomalies in large radio datasets while efficiently discarding millions of terrestrial signals, with the results described in a Nature Astronomy paper and a preprint available, the code on GitHub including data for the top candidates, and the full dataset accessible in the Open Data Archive, while background art is credited to Breakthrough Listen and Danielle Futselaar.
