BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//pretalx//pretalx.adass.org//adass2026//speaker//GNJ9DE
BEGIN:VTIMEZONE
TZID:AWST
BEGIN:STANDARD
DTSTART:20000101T000000
RRULE:FREQ=YEARLY;BYMONTH=1;UNTIL=20051231T160000Z
TZNAME:AWST
TZOFFSETFROM:+0800
TZOFFSETTO:+0800
END:STANDARD
BEGIN:STANDARD
DTSTART:20070325T040000
RRULE:FREQ=YEARLY;BYDAY=-1SU;BYMONTH=3
TZNAME:AWST
TZOFFSETFROM:+0900
TZOFFSETTO:+0800
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20071028T030000
RRULE:FREQ=YEARLY;BYDAY=4SU;BYMONTH=10;UNTIL=20081025T190000Z
TZNAME:AWDT
TZOFFSETFROM:+0800
TZOFFSETTO:+0900
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:pretalx-adass2026-MWJB3T@pretalx.adass.org
DTSTART;TZID=AWST:20261103T143000
DTEND;TZID=AWST:20261103T144500
DESCRIPTION:Radio interferometers and wide-area continuum surveys are produ
 cing increasingly large and complex imaging datasets\, making manual sourc
 e identification and classification impractical. Accurate\, fast and fully
  automated methods are therefore required to detect and characterise two d
 istinct but related populations: extended radio sources associated primari
 ly with active galactic nuclei\, including jets and radio lobes\, and diff
 use low-surface-brightness sources\, such as radio halos\, relics\, bridge
 s and filamentary emission associated with galaxy clusters and the cosmic 
 web. Both classes exhibit complex\, irregular and frequently multi-compone
 nt morphologies that are difficult to recover with conventional source-fin
 ding algorithms.\nTUNA (TransUNet for Astrophysical data) has been introdu
 ced as a deep-learning framework for the automated segmentation of radio s
 ources. Its effectiveness has already been investigated across all the tar
 geted source classes\, including extended radio galaxies and diffuse halos
 \, relics and bridges. TUNA is based on a hybrid TransUNet architecture co
 mbining a convolutional encoder\, a Vision Transformer and a U-Net-like de
 coder. Convolutional layers extract local brightness and morphological fea
 tures\, while multi-head self-attention captures long-range spatial depend
 encies between image regions. Skip connections preserve fine spatial infor
 mation during the reconstruction of full-resolution pixel-level segmentati
 on maps.\nThe TUNA training and inference pipelines are designed to exploi
 t GPU acceleration\, multi-GPU data parallelism and mixed-precision comput
 ation. Computationally intensive convolutional and attention operations ar
 e executed on GPUs\, while the workload can be distributed across multiple
  devices. Mixed-precision execution effects on memory and power consumptio
 n\, and computational performance is being investigated. \nThis contributi
 on will present the application of TUNA to the systematic processing of LO
 FAR Two-metre Sky Survey Data Release 3 pointings. The talk will describe 
 the processing workflow and assess the method from both scientific and com
 putational perspectives. Its effectiveness in detecting and segmenting ext
 ended and diffuse radio sources will be evaluated together with its comput
 ational efficiency\, scalability\, GPU utilisation and multi-GPU performan
 ce. Scientific accuracy and computational performance will be presented an
 d discussed as complementary requirements for the deployment of AI-based s
 ource-segmentation methods in current and future large-area radio surveys.
DTSTAMP:20261001T101837Z
LOCATION:Banquet Hall
SUMMARY:Scalable Deep-Learning Segmentation of Extended and Diffuse Radio S
 ources with TUNA - Claudio Gheller
URL:https://pretalx.adass.org/adass2026/talk/MWJB3T/
END:VEVENT
END:VCALENDAR