BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.2.3.1//EN
TZID:America/New_York
X-WR-TIMEZONE:America/New_York
BEGIN:VEVENT
UID:0-6355@eng.ufl.edu
DTSTART;TZID=America/New_York:20240123T120000
DTEND;TZID=America/New_York:20240123T130000
DTSTAMP:20251201T210800Z
URL:https://www.eng.ufl.edu/news-events/events/cise-faculty-candidate-semi
 nar-dr-xiao-fu/
SUMMARY:CISE Faculty Candidate Seminar: Dr. Xiao Fu
DESCRIPTION:Zoom Link: https://ufl.zoom.us/j/99350698347#success\nBio: Xiao
  Fu has been with the School of Electrical Engineering and Computer Scienc
 e\, Oregon State University since 2017\, where he is currently an Associat
 e Professor. He received the Ph.D. degree in Electronic Engineering from T
 he Chinese University of Hong Kong\, in 2014. He was a Postdoctoral Associ
 ate with the Department of Electrical and Computer Engineering\, Universit
 y of Minnesota - Twin Cities\, from 2014 to 2017. His research interests i
 nclude the broad area of machine learning and signal processing\, especial
 ly theory and algorithms.Dr. Fu received the Best Student Paper Award at I
 CASSP 2014\, the 2022 IEEE Signal Processing Society (SPS) Best Paper Awar
 d\, and the 2022 IEEE SPS Donald G. Fink Overview Paper Award. He also rec
 eived the Outstanding Postdoctoral Scholar Award at University of Minnesot
 a in 2016\, the Engelbrecht Early Career Faculty Award from the College of
  Engineering at Oregon State University in 2023\, and the National Science
  Foundation (NSF) CAREER Award in 2022. Since 2023\, he has been the Chair
  of the IEEE SPS Oregon Chapter. He is currently an Associate Editor of IE
 EE Transactions on Signal Processing. He was a tutorial speaker at ICASSP 
 2017 and SIAM Conference on Linear Algebra 2021.\nTitle: Towards Provable 
 Multimodal Learning: A Model Identification Perspective\nAbstract: 2023 wa
 s “the year of AI”\, highlighted by the release of numerous AI models 
 with remarkable capabilities. Multimodal learning is at the forefront of A
 I advancements\, with state-of-the-art models like GPT-4 and Gemini emphas
 izing multimodal functionalities as their defining features. Despite its i
 mportance\, many aspects of multimodal learning\, and AI developments in g
 eneral\, still lack a concrete and comprehensive understanding---which is 
 essential for building resilient and trustworthy systems. Our research foc
 uses on the understanding of AI/ML systems to drive theory-backed advancem
 ents. From this perspective\, this presentation revisits a core component 
 of multimodal learning—Unsupervised Domain Translation (UDT).Many UDT sy
 stems\, such as CycleGAN\, use Distribution Matching (DM) modules\, which 
 often fail in content-aligned translations due to measure-preserving autom
 orphism (MPA). Existing remedies fall short of guaranteed performance. In 
 my talk\, I will introduce a model identification perspective for UDT\, ov
 ercoming the MPA issues and ensuring identifiability of the desired transl
 ation functions. This is the first proven identification result in UDT und
 er CycleGAN’s settings\, to our knowledge. We have also broadened these 
 concepts\, providing solutions for various translation challenges\, enabli
 ng provable content-style disentanglement\, and offering more versatile cr
 oss-domain data generation. These advancements promise significant theoret
 ically supported enhancements for UDT applications\, particularly in data-
 limited fields such as medicine and biology.
CATEGORIES:Seminars
LOCATION:Nvidia Auditorium 1000\, 1889 Museum Road\, Gainesville\, Florida\
 , 32611\, United States
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=1889 Museum Road\, Gainesvi
 lle\, Florida\, 32611\, United States;X-APPLE-RADIUS=100;X-TITLE=Nvidia Au
 ditorium 1000:geo:0,0
END:VEVENT
BEGIN:VTIMEZONE
TZID:America/New_York
X-LIC-LOCATION:America/New_York
BEGIN:STANDARD
DTSTART:20231105T010000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
TZNAME:EST
END:STANDARD
END:VTIMEZONE
END:VCALENDAR