BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Denver
X-LIC-LOCATION:America/Denver
BEGIN:DAYLIGHT
TZOFFSETFROM:-0700
TZOFFSETTO:-0600
TZNAME:MDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0600
TZOFFSETTO:-0700
TZNAME:MST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20260422T000711Z
LOCATION:501-502
DTSTART;TZID=America/Denver:20231113T164500
DTEND;TZID=America/Denver:20231113T170000
UID:submissions.supercomputing.org_SC23_sess440_ws_ai4s103@linklings.com
SUMMARY:Enhancing Heterogeneous Federated Learning with Knowledge Extracti
 on and Multi-Model Fusion
DESCRIPTION:Duy Phuong Nguyen and Sixing Yu (Iowa State University), J. Pa
 blo Muñoz (Intel Corporation), and Ali Jannesari (Iowa State University)\n
 \nConcerned with user data privacy, this paper presents a new federated le
 arning (FL) method that trains machine learning models on edge devices wit
 hout accessing sensitive data. Traditional FL methods, although privacy-pr
 otective, fail to manage model heterogeneity and incur high communication 
 costs due to their reliance on aggregation methods. To address this limita
 tion, we propose a resource-aware FL method that aggregates local knowledg
 e from edge models and distills it into robust global knowledge through kn
 owledge distillation. This method allows efficient multi-model knowledge f
 usion and the deployment of resource-aware models while preserving model h
 eterogeneity. Our method improves communication cost and performance in he
 terogeneous data and models compared to existing FL algorithms. Notably, i
 t reduces the communication cost of ResNet-32 by up to 50% and VGG-11 by u
 p to 10x while delivering superior performance.\n\nTag: Artificial Intelli
 gence/Machine Learning\n\nRegistration Category: Workshop Reg Pass\n\nSess
 ion Chairs: Murali Emani (Argonne National Laboratory (ANL)); Gokcen Kesto
 r (Barcelona Supercomputing Center (BSC); University of California, Merced
 ); and Dong Li (University of California, Merced)\n\n
END:VEVENT
END:VCALENDAR