BEGIN:VCALENDAR
VERSION:2.0
PRODID:IEEE vTools.Events//EN
CALSCALE:GREGORIAN
BEGIN:VTIMEZONE
TZID:America/New_York
BEGIN:DAYLIGHT
DTSTART:20220313T030000
TZOFFSETFROM:-0500
TZOFFSETTO:-0400
RRULE:FREQ=YEARLY;BYDAY=2SU;BYMONTH=3
TZNAME:EDT
END:DAYLIGHT
BEGIN:STANDARD
DTSTART:20211107T010000
TZOFFSETFROM:-0400
TZOFFSETTO:-0500
RRULE:FREQ=YEARLY;BYDAY=1SU;BYMONTH=11
TZNAME:EST
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
DTSTAMP:20211227T195536Z
UID:8D16BD94-D974-4A12-98D5-51851837F5FD
DTSTART;TZID=America/New_York:20211227T113000
DTEND;TZID=America/New_York:20211227T123000
DESCRIPTION:Please join us for a presentation by Dr. Walter Bennette on Hie
 rarchical Open-Set Recognition for Automatic Target Recognition!\n\nAbstra
 ct: Traditional classification and recognition models are trained with an 
 assumption that all possible classes are encountered during model training
 . When such models operate in the open-world\, they can encounter observat
 ions from previously unknown classes\, and are prone to making erroneous h
 igh confidence predictions. In this work\, we define the problem of hierar
 chical Open-Set Recognition (OSR) which seeks classification models that a
 re able to identify if an observation belongs to an unknown class\, and to
  provide information about how the unknown class relates to the known clas
 ses. As part of this work we introduce three novel approaches to hierarchi
 cal OSR\, and a novel evaluation metric that conveys the ability of hierar
 chical OSR models to provide partial information for predictions. Through 
 empirical results we demonstrate that one of our novel hierarchical OSR ap
 proaches outperforms competing methods. Finally\, we introduce a use case 
 that shows the benefit of hierarchical OSR for Automatic Target Recognitio
 n applications.\n\nSpeaker(s): Dr. Walter Bennette\, \n\nVirtual: https://
 events.vtools.ieee.org/m/295655
LOCATION:Virtual: https://events.vtools.ieee.org/m/295655
ORGANIZER:ashley.prater@ieee.org
SEQUENCE:3
SUMMARY:Mohawk Valley Section AES Technical Presentation - Hierarchical Ope
 n-Set Recognition for Automatic Target Recognition
URL;VALUE=URI:https://events.vtools.ieee.org/m/295655
X-ALT-DESC:Description: <br /><p>Please join us for a presentation by Dr. W
 alter Bennette on Hierarchical Open-Set Recognition for Automatic Target R
 ecognition!</p>\n<p>Abstract:&nbsp\;Traditional classification and recogni
 tion models are trained with an&nbsp\;assumption that all possible classes
  are encountered during model training.&nbsp\;When such models operate in 
 the open-world\, they can encounter observations&nbsp\;from previously unk
 nown classes\, and are prone to making erroneous high&nbsp\;confidence pre
 dictions. In this work\, we define the problem of hierarchical&nbsp\;Open-
 Set Recognition (OSR) which seeks classification models that are able&nbsp
 \;to identify if an observation belongs to an unknown class\, and to provi
 de&nbsp\;information about how the unknown class relates to the known clas
 ses. As&nbsp\;part of this work we introduce three novel approaches to&nbs
 p\; hierarchical OSR\,&nbsp\;and a novel evaluation metric that conveys th
 e ability of hierarchical OSR&nbsp\;models to provide partial information 
 for predictions. Through empirical&nbsp\;results we demonstrate that one o
 f our novel hierarchical OSR approaches&nbsp\;outperforms competing method
 s. Finally\, we introduce a use case that shows&nbsp\;the benefit of&nbsp\
 ;hierarchical OSR for Automatic Target Recognition applications.</p>
END:VEVENT
END:VCALENDAR