CANCELLED: POSTER P407: WALLABY: Source-finding at lower signal-to-noise using a machine learning framework

pretalx

AI in Astronomy

We introduce a novel deep learning framework for detecting low Signal-to-Noise Ratio (SNR) HI sources with enhanced reliability from the ASKAP WALLABY wide-area survey. Leveraging the spatial and depth capabilities of 3D Convolutional Neural Networks (CNNs), our method is specifically designed to recognize patterns and features in three-dimensional space, making it uniquely suited for detecting sources in low SNR scenarios that might be missed by conventional linear methods. As a result, our approach is significantly more accurate in source detection and results in considerably fewer false detections compared to previous linear-based source finders. Performance tests using mock galaxies injected into genuine ASKAP data reveal our method's capability to achieve near-100% completeness and reliability at a relatively low integrated SNR of about 3-5. Moreover, our approach, trained using data from one sky region, consistently showed robust performance when tested in other sky areas, underscoring its high adaptability. An at-scale version of this tool will greatly maximise the science output from the upcoming widefield HI surveys.