Mitigating Radio Frequency Interference in Radio Astronomy: Recent Progress at Brown

Radio frequency interference (RFI) is a pressing challenge for radio astronomy, particularly for experiments targeting the Cosmic Dark Ages and the Epoch of Reionization. As the number of artificial radio emitters continues to grow, RFI has become increasingly difficult to avoid. Significant research effort is therefore dedicated to identifying, characterizing, and flagging RFI in interferometric data. In this talk, I will present a brief overview of ongoing RFI mitigation research at Brown. First, I will describe a machine-learning-based RFI classification approach based on gradient-boosted decision trees. The model is trained in a weak-supervision framework that combines labels from five independent flagging algorithms, each applied to over 20 TB of data from the Murchison Widefield Array. Preliminary results are encouraging. I will then introduce a method for estimating the altitude of RFI-emitting sources in interferometric data, allowing a full three-dimensional localization of the source and representing a first step towards RFI subtraction as opposed to mere flagging. Together, these approaches address RFI mitigation from both a data-driven and a physically motivated perspective.

Biography: Jade Ducharme is a fourth-year PhD student in Physics at Brown University. Her research focuses on characterizing and mitigating radio-frequency interference in interferometric radio data, with an emphasis on observations from the Murchison Widefield Array.