A deep-learning search for technosignatures from 820 nearby stars

Abstract

The goal of the Search for Extraterrestrial Intelligence (SETI) is to quantify the prevalence of technological life beyond Earth via their "technosignatures". One theorized technosignature are narrowband Doppler drifting radio signals. The principal challenge in conducting SETI in the radio domain is developing a generalized technique to reject human radio frequency interference (RFI) that dominate the features across the band in searches for technosignatures. Here, we present the first comprehensive deep-learning based technosignature search to date, returning 8 promising ETI signals-of-interest for re-observation as part of the Breakthrough Listen initiative. The search comprises 820 unique targets observed with the Robert C. Byrd Green Bank Telescope, totaling over 480 hr of on-sky data. We implement a novel β −Convolutional Variational Autoencoder with an embedded discriminator combined with Random Forest Decision Trees to classify technosignature candidates in a semiunsupervised manner. We compare our results with prior classical techniques on the same dataset and conclude that our algorithm returns more convincing and novel signals-of-interest with a manageable false positive rate. This new approach presents itself as a leading solution in accelerating SETI and other transient research into the age of data-driven astronomy.

On Detecting Interstellar Scintillation in Narrowband Radio SETI

“…Machine learning (ML) could be used to aid scintillated searches, such as for creating initial classifications of signal type and eventually even for doing final candidate analysis. In particular, deep learning techniques, such as convolutional neural networks (CNNs), have been used effectively in a variety of tasks using radio spectrograms (Zhang et al 2018;Harp et al 2019;Brzycki et al 2020;Pinchuk & Margot 2022;Ma et al 2023). CNNs could be used to filter out spectrograms with clear broadband emission and would be relatively straightforward to integrate into the pipeline.…”

AI becomes a masterbrain scientist

“…Over the past few years, with the rapid development of AI technologies such as deep learning [12], AI has been widely applied in various scientific fields, driving significant advancements in areas such as medicine and life sciences [13][14][15], drug discovery [16,17], protein structure prediction and design [18,19], chemistry [20,21], materials science [22,23], physics [24,25], astronomy [26,27], energy science [28,29], environmental science [30,31], and many others.…”

The Breakthrough Listen Search for Intelligent Life: Technosignature Search of 97 Nearby Galaxies

“…Price et al (2020) proposed to update the turboSETI algorithm to employ a drift-dependent moving boxcar, averaging N adjacent channels of the integrated spectra to recover smeared power. Future work will implement some of these solutions, and alternative avenues, such as machine learning, are also being explored (Ma et al 2023), as well as more computationally intensive overlapping channelization schemes to combat the loss of power due to spectral leakage. At present, no windowing function is applied during the fine channelization of BL data products; doing so could potentially improve channel isolation and reduce spectral leakage.…”