Optimizing sparse RFI prediction using deep learning

TLDR

Radio Frequency Interference limits sensitivity of radio telescopes, especially for Epoch of Reionization studies, and efficient algorithms are needed as arrays like HERA grow. The authors propose a Deep Fully Convolutional Neural Network that jointly uses amplitude and phase interferometric data to identify RFI. The network is trained on simulated HERA visibilities with mock RFI, providing a ground‑truth dataset for evaluating algorithm accuracy. On HERA‑67 data, the DFCN processes 1.6 × 10⁵ time‑ordered visibilities per hour per GPU, and using phase improves discrimination, achieving recall 0.81, precision 0.58, and F₂ score 0.75.

Abstract

Radio Frequency Interference (RFI) is an ever-present limiting factor among radio telescopes even in the most remote observing locations. When looking to retain the maximum amount of sensitivity and reduce contamination for Epoch of Reionization studies, the identification and removal of RFI is especially important. In addition to improved RFI identification, we must also take into account computational efficiency of the RFI-Identification algorithm as radio interferometer arrays such as the Hydrogen Epoch of Reionization Array grow larger in number of receivers. To address this, we present a Deep Fully Convolutional Neural Network (DFCN) that is comprehensive in its use of interferometric data, where both amplitude and phase information are used jointly for identifying RFI. We train the network using simulated HERA visibilities containing mock RFI, yielding a known "ground truth" dataset for evaluating the accuracy of various RFI algorithms. Evaluation of the DFCN model is performed on observations from the 67 dish build-out, HERA-67, and achieves a data throughput of 1.6$\times 10^{5}$ HERA time-ordered 1024 channeled visibilities per hour per GPU. We determine that relative to an amplitude only network including visibility phase adds important adjacent time-frequency context which increases discrimination between RFI and Non-RFI. The inclusion of phase when predicting achieves a Recall of 0.81, Precision of 0.58, and $F_{2}$ score of 0.75 as applied to our HERA-67 observations.