Stephen Bach, Jonathan Pober, Bryna Hazelton, Miguel Morales, Alexander Ratner, Shawn Dubey, Jade Ducharme, Charles Duong, Aidan LaBella, Yishan Liu, Christopher Tripp, Eli Lilleskov


This poster provides updates on the SWIFT-SAT project "RFI Detection Across Six Orders of Magnitude in Intensity: A Unifying Framework with Weakly Supervised Machine Learning." We report progress in several areas:
– Developed a framework to reliably train U-Net models on MWA data using non-ML detection method ground truths.
– Developed a method to evaluate trained U-Net model performance by comparison to non-ML method, using metrics such as the intersection-over-union.
– Began development of new methods to identify RFI sources in image space.
– Developed a new near-field interferometry method for locating and imaging local RFI sources in MWA data.