BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:America/Bogota
BEGIN:DAYLIGHT
DTSTART:20380118T221407
TZOFFSETFROM:-0500
TZOFFSETTO:-0500
RRULE:FREQ=YEARLY;BYDAY=3MO;BYMONTH=1
TZNAME:-05
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:19930206T230000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
RRULE:FREQ=YEARLY;BYDAY=1SA;BYMONTH=2
TZNAME:-05
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20231002T183557Z
UID:D135670F-9EEB-44E6-947A-C7C4726BFC3E
DTSTART;TZID=America/Bogota:20230821T130000
DTEND;TZID=America/Bogota:20230821T150000
DESCRIPTION:In this talk we discuss approaches for distributed machine lear
 ning (ML) in wireless networks\, in particular vehicular networks and reso
 urce-constrained Internet of Things (IoT) networks. Federated Learning (FL
 ) and Split Learning (FL) are popular approaches in such environments.\nFi
 rst\, we present Early Exit of Communication (EEoC)\, which adaptively spl
 its ML inference in an IoT edge computing environment to meet latency and 
 energy constraints. This layer-based (vertically partitioned) approach has
  been extended by Distributed Micro-Split Deep Learning in Heterogeneous D
 ynamic IoT (DISNET)\, which adds horizontal partitioning to better support
  flexible\, distributed\, and parallel execution of neural network models 
 on heterogeneous IoT devices under dynamic conditions. Then\, we also cons
 ider the training aspect by developing and evaluating Adaptive REsource-aw
 are Split-learning (ARES)\, a scheme for efficient model training in IoT s
 ystems. Recent work suggests Dynamic FL (DFL) for heterogeneous IoT\, whic
 h uses resource-aware SL and FL based on similarity-based layer-wise model
  aggregation.\n\nCo-sponsored by: Universidad Tecnológica de Panamá\n\nS
 peaker(s): Torsten Braun\n\nAgenda: \nRegistro de participantes\n\nConfere
 ncia\n\nPreguntas y respuestas\n\nDespedida\n\nRoom: Salón Roberto Barraz
 a\, Bldg: Edificio 1\, Salón Roberto Barraza\, Universidad Tecnológica d
 e Panamá\, Campus Víctor Levi Sasso\, Panama\, Panama\, Panama
LOCATION:Room: Salón Roberto Barraza\, Bldg: Edificio 1\, Salón Roberto B
 arraza\, Universidad Tecnológica de Panamá\, Campus Víctor Levi Sasso\,
  Panama\, Panama\, Panama
ORGANIZER:aris.castillo@utp.ac.pa
SEQUENCE:31
SUMMARY:Distributed Machine Learning in IoT and Vehicular Networks
URL;VALUE=URI:https://events.vtools.ieee.org/m/369060
X-ALT-DESC:Description: <br /><p>In this talk we discuss approaches for dis
 tributed machine learning (ML) in wireless networks\, in particular vehicu
 lar networks and resource-constrained Internet of Things (IoT) networks. F
 ederated Learning (FL) and Split Learning (FL) are popular approaches in s
 uch environments.&nbsp\;<br aria-hidden="true" />First\, we present Early 
 Exit of Communication (EEoC)\, which adaptively splits ML inference in an 
 IoT edge computing environment to meet latency and energy constraints. Thi
 s layer-based (vertically partitioned) approach has been extended by Distr
 ibuted Micro-Split Deep Learning in Heterogeneous Dynamic IoT (DISNET)\, w
 hich adds horizontal partitioning to better support flexible\, distributed
 \, and parallel execution of neural network models on heterogeneous IoT de
 vices under dynamic conditions. Then\, we also consider the training aspec
 t by developing and evaluating Adaptive REsource-aware Split-learning (ARE
 S)\, a scheme for efficient model training in IoT systems. Recent work sug
 gests Dynamic FL (DFL) for heterogeneous IoT\, which uses resource-aware S
 L and FL based on similarity-based layer-wise model aggregation.</p><br />
 <br />Agenda: <br /><p>Registro de participantes</p>\n<p>Conferencia</p>\n
 <p>Preguntas y respuestas</p>\n<p>Despedida</p>
END:VEVENT
END:VCALENDAR