BEGIN:VCALENDAR VERSION:2.0 PRODID:IEEE vTools.Events//EN CALSCALE:GREGORIAN BEGIN:VTIMEZONE TZID:America/Chicago BEGIN:DAYLIGHT DTSTART:20250309T030000 TZOFFSETFROM:-0600 TZOFFSETTO:-0500 RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3 TZNAME:CDT END:DAYLIGHT BEGIN:STANDARD DTSTART:20251102T010000 TZOFFSETFROM:-0500 TZOFFSETTO:-0600 RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11 TZNAME:CST END:STANDARD END:VTIMEZONE BEGIN:VEVENT DTSTAMP:20250522T035100Z UID:3CEF35BD-2AEC-4B5D-9F5E-46F6066B510F DTSTART;TZID=America/Chicago:20250521T090000 DTEND;TZID=America/Chicago:20250521T100000 DESCRIPTION:This talk explores the intersection of Deep Learning\, Generati ve AI\, and Federated Learning in advancing solutions for Multimedia Proce ssing\, Human-Computer Interaction (HCI)\, and Medical Healthcare Applicat ions. The session aims to present a set of intelligent frameworks that add ress real-world challenges through scalable\, privacy-aware\, and high-per formance learning systems.\nBeginning with biomedical signal analysis\, th e talk introduces LAACNN\, a hybrid architecture combining convolutional\, recurrent\, and attention mechanisms for classifying neuromuscular diseas es using raw EMG signals. This approach enhances early diagnostic potentia l through non-invasive\, data-driven insights.\nIn the realm of advanced m ultimedia\, speech enhancement is tackled through two novel Generative AI models. CAR-UNet employs ConvNeXT blocks and gated attention to jointly en hance magnitude and phase information\, ensuring clarity under extreme noi se conditions. N2N2N\, a clean-data-independent approach using a modified cGAN\, refines noisy speech where clean references are unavailable\, makin g it suitable for realistic deployment.\nFor Human-Computer Interaction\, the session features CAPRes50-GAN\, a real-time\, GAN-based classifier for word-level sign language recognition. It leverages multi-head attention a longside ResNet-based architectures to efficiently interpret gestures and support inclusive communication systems.\nThe discussion continues with Br eastDCGAN\, an end-to-end framework for breast cancer detection using mamm ography and ultrasound. It combines generative segmentation and attention- based classification to assist in accurate and early medical analysis.\nCo ncluding the talk\, pFLOCT is introduced—a Personalized Federated Learni ng framework for optical imaging that accommodates heterogeneous client da ta while preserving privacy\, advancing collaborative learning in medical domains.\n\nSpeaker(s): Dr. Celia Shahnaz\, Dept. of EEE\, BUET.\n\nVirtua l: https://events.vtools.ieee.org/m/485235 LOCATION:Virtual: https://events.vtools.ieee.org/m/485235 ORGANIZER:manishaguduri@ieee.org SEQUENCE:118 SUMMARY:Summer Digital Dialogues: Deep Learning\, Generative AI and Federat ed Learning for Advanced Multimedia\, Human Computer Interface and Biomedi cal Applications URL;VALUE=URI:https://events.vtools.ieee.org/m/485235 X-ALT-DESC:Description: <br /><p style="text-align: justify\;">This talk ex plores the intersection of Deep Learning\, Generative AI\, and Federated L earning in advancing solutions for Multimedia Processing\, Human-Computer Interaction (HCI)\, and Medical Healthcare Applications. The session aims to present a set of intelligent frameworks that address real-world challen ges through scalable\, privacy-aware\, and high-performance learning syste ms.<br>Beginning with biomedical signal analysis\, the talk introduces LAA CNN\, a hybrid architecture combining convolutional\, recurrent\, and atte ntion mechanisms for classifying neuromuscular diseases using raw EMG sign als. This approach enhances early diagnostic potential through non-invasiv e\, data-driven insights.<br>In the realm of advanced multimedia\, speech enhancement is tackled through two novel Generative AI models. CAR-UNet em ploys ConvNeXT blocks and gated attention to jointly enhance magnitude and phase information\, ensuring clarity under extreme noise conditions. N2N2 N\, a clean-data-independent approach using a modified cGAN\, refines nois y speech where clean references are unavailable\, making it suitable for r ealistic deployment.<br>For Human-Computer Interaction\, the session featu res CAPRes50-GAN\, a real-time\, GAN-based classifier for word-level sign language recognition. It leverages multi-head attention alongside ResNet-b ased architectures to efficiently interpret gestures and support inclusive communication systems.<br>The discussion continues with BreastDCGAN\, an end-to-end framework for breast cancer detection using mammography and ult rasound. It combines generative segmentation and attention-based classific ation to assist in accurate and early medical analysis.<br>Concluding the talk\, pFLOCT is introduced&mdash\;a Personalized Federated Learning frame work for optical imaging that accommodates heterogeneous client data while preserving privacy\, advancing collaborative learning in medical domains. </p> END:VEVENT END:VCALENDAR