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:20240626T180034Z
LOCATION:Level 2 Lobby
DTSTART;TZID=America/Los_Angeles:20240626T180000
DTEND;TZID=America/Los_Angeles:20240626T190000
UID:dac_DAC 2024_sess256_LBR120@linklings.com
SUMMARY:Late Breaking Results: Evaluation of Human Action Quality with Lin
 ear Recurrent Units and Graph Attention Networks on Embedded Systems.
DESCRIPTION:Late Breaking Results Poster\n\nFilippo Ziche and Nicola Bombi
 eri (University of Verona)\n\nRecent evolutions of recurrent neural networ
 ks (RNN) such as S4, S4D, and LRU, have shown remarkable potential for ver
 y long-range sequence modeling tasks for vision, language, and audio. They
  have shown a capacity to capture dependencies over tens of thousands of s
 teps. Unlike transformers, which face significant memory consumption chall
 enges with large context sizes, they are a promising alternative with thei
 r ability to operate effectively on embedded systems. While they have been
  evaluated for classification and segmentation tasks, no work in the liter
 ature has applied them in the context of human pose estimation. In this wo
 rk we propose an architecture that combines such state space models (SSM) 
 to graph attention networks (GAT) to enable their application to evaluate 
 human action tasks on embedded systems.
END:VEVENT
END:VCALENDAR