MROP: A Novel Data Dimensionality Reduction Scheme for Imaging in Radio Astronomy
Find out more about subscribing to add all events.
MROP: A Novel Data Dimensionality Reduction Scheme for Imaging in Radio Astronomy
Fri, 24/10/2025 - 14:00 to 14:30
Location:
Speaker:
Taylor Chu
Affiliation:
Heriot Watt
Synopsis:
The emerging generation of radio-interferometric (RI) arrays, such as the upcoming Square Kilometre Array (SKA), are set to observe the sky with unprecedented angular resolution and sensitivity. This ambition leads to the requirement of arrays with a large number of antennas. As the RI data volume scales with the number of snapshots and quadratically with the number of antennas, Exabyte-scale data volumes are to be expected for arrays such as SKA, calling for efficient data dimensionality reduction techniques.
The Modulated Rank-One Projection (MROP) model is a new approach to RI data compression during acquisition. The talk will cover MROP’s foundation, its validation on simulated single-frequency RI data, and a comparison of image reconstruction against both uncompressed data and baseline-dependent averaging (BDA) models. Finally, this talk will highlight MROP’s validation on multi-frequency real RI data, demonstrating its potential for large-scale radio astronomy applications.
Biography:
Taylor Chu is a final year PhD student in the BASP group at ISSS, Heriot-Watt University, supervised by Prof. Yves Wiaux. He is a member of the 2022 cohort of the MAC-MIGS CDT, jointly run by Heriot-Watt University and University of Edinburgh. He earned his MSc in Applied Data Science from Heriot-Watt University in 2022 and his MEng in Aeronautical Engineering from Imperial College London in 2017. His research focuses on deep learning-based imaging algorithms for radio astronomy, with applications including imaging the M87 supermassive blackhole observed by the Event Horizon Telescope (EHT), and developing data dimensionality reduction schemes for next-generation radio interferometers.
Website: