BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:Australia/Sydney BEGIN:DAYLIGHT DTSTART:20261004T030000 TZOFFSETFROM:+1000 TZOFFSETTO:+1100 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=10 TZNAME:AEDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20270404T020000 TZOFFSETFROM:+1100 TZOFFSETTO:+1000 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=4 TZNAME:AEST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20261008T025756Z UID:30DB9D5C-DDA3-4723-918F-A180B33112EC DTSTART;TZID=Australia/Sydney:20261028T173000 DTEND;TZID=Australia/Sydney:20261028T183000 DESCRIPTION:Join the IEEE New South Wales Signal Processing Society Chapter for an online seminar by Professor Manoranjan Paul from Charles Sturt Uni versity.\n\nDate: Wednesday\, 28 October 2026\nTime: 5:30–6:30 pm AEDT ( Sydney time)\nVenue: Online via Zoom:\n\nLINK: https://cqu.zoom.us/j/85106 607089?pwd=mUcbbLO9bawrKvdNsMnaG1ainQEzwO.1\n\nSpeaker Profile: Professor Manoranjan Paul\n\nProfessor Manoranjan Paul is a Professor of Computer Sc ience at Charles Sturt University\, Director of its Computer Vision Lab\, and Head of the Machine Vision and Digital Health Research Group. An award -winning researcher\, he received the ACS ICT Researcher of the Year and G olden Disruptor awards in 2017. He has published around 200 peer-reviewed publications\, supervised 16 PhD completions\, and secured $3.6 million in competitive external research funding. He serves as an Associate Editor f or leading IEEE and EURASIP journals and is a Senior Member of IEEE and AC S.\n\nTitle: AI That Thinks Differently: Seeing\, Understanding\, and Trus ting Visual Data in the Age of Foundation Models\n\nVisual data now underp ins healthcare\, agriculture\, autonomous systems and everyday communicati on\, yet it remains costly to store and transmit\, difficult to interpret at a fine-grained level\, and increasingly easy to fake. This talk present s our group's research on AI that thinks differently about how machines pe rceive\, represent and reason over visual content. Drawing on years of wor k in digital health\, environmental sensing and intelligent systems\, I sh ow that domain knowledge\, structure-aware design and careful handling of noisy\, real-world data are what turn accurate models into reliable\, depl oyable ones. Building on this foundation\, I present our recent efforts al ong four connected themes: efficient representation\, where a "compress on ce\, adapt many" paradigm allows a single compact encoding of video and 3D data to serve many downstream machine vision tasks\; fine-grained underst anding\, where vision-language models are guided to link visual parts with semantic descriptions\, even with few examples\; spatio-temporal reasonin g\, where compact\, task-aware video representations support efficient que stion answering\; and authenticity\, where structural cues are used to exp ose the subtle traces left by generative models in fake images. Together\, these themes outline a path toward visual AI that is efficient\, discrimi native and trustworthy. I conclude with open challenges and opportunities for collaboration\, including unified multimodal pipelines and emerging co mputing paradigms for imaging.\n\nSpeaker(s): Professor Manoranjan Paul \n \nAgenda: \n5:30 pm — Welcome and IEEE SPS Chapter and speaker introduct ions\n5:35 pm — Presentation by Professor Manoranjan Paul\n6:15 pm — Q uestions and discussion\n6:30 pm — Close\n\nVirtual: https://events.vtoo ls.ieee.org/m/581284 LOCATION:Virtual: https://events.vtools.ieee.org/m/581284 ORGANIZER:a.anwaarulhaq@cqu.edu.au SEQUENCE:52 SUMMARY:IEEE SPS NSW Webinar Series 2026–27: Inaugural Keynote Talk URL;VALUE=URI:https://events.vtools.ieee.org/m/581284 X-ALT-DESC:Description: <br /><p data-pm-slice="1 1 []">Join the IEEE New S outh Wales Signal Processing Society Chapter for an online seminar by Prof essor Manoranjan Paul from Charles Sturt University.</p>\n<p data-pm-slice ="1 1 []">Date: Wednesday\, 28 October 2026<br>Time: 5:30&ndash\;6:30 pm A EDT (Sydney time)<br>Venue: Online via Zoom:&nbsp\;</p>\n<p data-pm-slice= "1 1 []">LINK: <a href="https://cqu.zoom.us/j/85106607089?pwd=mUcbbLO9bawr KvdNsMnaG1ainQEzwO.1">https://cqu.zoom.us/j/85106607089?pwd=mUcbbLO9bawrKv dNsMnaG1ainQEzwO.1</a></p>\n<p class="MsoNormal"><strong>Speaker Profile: Professor Manoranjan Paul&nbsp\;</strong></p>\n<p class="MsoNormal">Profes sor Manoranjan Paul is a Professor of Computer Science at Charles Sturt Un iversity\, Director of its Computer Vision Lab\, and Head of the Machine V ision and Digital Health Research Group. An award-winning researcher\, he received the ACS ICT Researcher of the Year and Golden Disruptor awards in 2017. He has published around 200 peer-reviewed publications\, supervised 16 PhD completions\, and secured $3.6 million in competitive external res earch funding. He serves as an Associate Editor for leading IEEE and EURAS IP journals and is a Senior Member of IEEE and ACS.<strong>&nbsp\;</strong ></p>\n<p class="MsoNormal"><strong>Title: AI That Thinks Differently: See ing\, Understanding\, and Trusting Visual Data in the Age of Foundation Mo dels</strong></p>\n<p class="MsoNormal">Visual data now underpins healthca re\, agriculture\, autonomous systems and everyday communication\, yet it remains costly to store and transmit\, difficult to interpret at a fine-gr ained level\, and increasingly easy to fake. This talk presents our group' s research on AI that thinks differently about how machines perceive\, rep resent and reason over visual content. Drawing on years of work in digital health\, environmental sensing and intelligent systems\, I show that doma in knowledge\, structure-aware design and careful handling of noisy\, real -world data are what turn accurate models into reliable\, deployable ones. Building on this foundation\, I present our recent efforts along four con nected themes: efficient representation\, where a "compress once\, adapt m any" paradigm allows a single compact encoding of video and 3D data to ser ve many downstream machine vision tasks\; fine-grained understanding\, whe re vision-language models are guided to link visual parts with semantic de scriptions\, even with few examples\; spatio-temporal reasoning\, where co mpact\, task-aware video representations support efficient question answer ing\; and authenticity\, where structural cues are used to expose the subt le traces left by generative models in fake images. Together\, these theme s outline a path toward visual AI that is efficient\, discriminative and t rustworthy. I conclude with open challenges and opportunities for collabor ation\, including unified multimodal pipelines and emerging computing para digms for imaging.</p><br /><br />Agenda: <br /><p data-pm-slice="1 1 []"> 5:30 pm &mdash\; Welcome and IEEE SPS Chapter and speaker introductions<br >5:35 pm &mdash\; Presentation by Professor Manoranjan Paul<br>6:15 pm &md ash\; Questions and discussion<br>6:30 pm &mdash\; Close</p> END:VEVENT END:VCALENDAR