BEGIN:VCALENDAR
VERSION:2.0
PRODID:Linklings LLC
BEGIN:VTIMEZONE
TZID:America/Los_Angeles
X-LIC-LOCATION:America/Los_Angeles
BEGIN:DAYLIGHT
TZOFFSETFROM:-0800
TZOFFSETTO:-0700
TZNAME:PDT
DTSTART:19700308T020000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=2SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:-0700
TZOFFSETTO:-0800
TZNAME:PST
DTSTART:19701101T020000
RRULE:FREQ=YEARLY;BYMONTH=11;BYDAY=1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20211208T164412Z
LOCATION:Level 2 - Lobby
DTSTART;TZID=America/Los_Angeles:20211208T180000
DTEND;TZID=America/Los_Angeles:20211208T190000
UID:dac_DAC 2021_sess243_WIP194@linklings.com
SUMMARY:A Coordinated GPU Overdrive Fault Attack on Neural Networks
DESCRIPTION:Networking Reception, Work-in-Progress Poster\n\nA Coordinated
  GPU Overdrive Fault Attack on Neural Networks\n\nKaeli\n\nGraphics proces
 sing units (GPUs) are commonly used to accelerate training and inference o
 f convolutional neural networks (CNNs). Modern GPUs can be shared by multi
 ple users to execute workloads concurrently. However, the reliability and 
 security of shared GPU computing have not been carefully evaluated. In thi
 s paper, we design a coordinated overdrive fault attack on GPUs. An advers
 ary kernel utilizes the shared GPU caches to detect the presence of a vict
 im kernel to trigger fault injections in time. We thoroughly characterize 
 the fault injections and propagation in a victim CNN. We successfully laun
 ched an end-to-end misclassification attack during CNN inference.\n\nVirtu
 al Program: Presented In-Person
END:VEVENT
END:VCALENDAR