BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:IEM Seminar series: Reliable  Real-Time Distributed AI for Mobile 
 Autonomous Systems
DTSTART:20220609T110000
DTEND:20220609T120000
DTSTAMP:20260609T201338Z
UID:692dca4538c495a602a79dc774e6bd55b875aeabfd324045fe2ee552
CATEGORIES:Conferences - Seminars
DESCRIPTION:Marco Levorato\, University of California\, Irvine\n\nAbstract
 : The autonomous operations of Unmanned Aerial Vehicles (UAV) require the 
 execution of continuous streams of heavy-duty mission-critical computing t
 asks. These tasks often take the form of complex Deep Neural Network (DNN)
  models applied to information-rich signals such as image and lidar data. 
 Clearly\, such strenuous effort may exceed the capabilities and resources 
 (e.g.\, computing power and energy) of most UAVs. The research community m
 ostly relied on two distinct approaches to address this issue: model simpl
 ification and edge computing. The former may lead to performance degradati
 on\, while the performance of the latter suffer from the erratic behavior 
 of wireless channels.\nIn this talk\, I will present two key components of
  reliable computing in the edge computing for Mobile Autonomous Systems (M
 AS) and especially UAVs. Our ultimate objective is to achieve task-level l
 atency and accuracy guarantees. First\, I will introduce the notion of dat
 a-driven redundant computing. The core idea is to replicate tasks across t
 he system to reduce uncertainty in their total execution time. A Deep Rein
 forcement Learning (DRL) Agent dynamically controls in real-time how many 
 and which edge servers are selected for computing. Our approach is experim
 ental\, and we demonstrate how the DRL agent necessitates features from mu
 ltiple system blocks (telemetry\, network and application) to make effecti
 ve decisions in real-world deployments. Finally\, I will summarize our wor
 k in the area of split computing\, where we modify DNN models for vision t
 o make them splittable across MAS and edge servers and reduce end-to-end l
 atency in unreliable wireless environments. We pioneered this area by intr
 oducing the notion of artificial bottleneck to obtain in-model compression
 \, and by developing innovative training strategies that achieve the best 
 rate distortion curve available to date.\n \n\n\n\nBio: Marco Levorato is
  an Associate Professor in the Computer Science department at UC Irvine. H
 e completed the PhD in Electrical Engineering at the University of Padova\
 , Italy\, in 2009. Between 2010 and 2012\, he was a postdoctoral researche
 r with a joint affiliation at Stanford and the University of Southern Cali
 fornia. His research interests are focused on distributed computing over u
 nreliable wireless systems\, especially for autonomous vehicles and health
 care systems. His work received the best paper award at IEEE GLOBECOM (201
 2). In 2016 and 2019\, he received the UC Hellman Foundation Award and the
  Dean mid-career research award\, respectively. His research is funded by 
 the National Science Foundation\, the Department of Defense\, Intel and Ci
 sco.\n
LOCATION:ELA 1 https://plan.epfl.ch/?room==ELA%201 https://epfl.zoom.us/j/
 65062056203
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR