BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Date iCal//NONSGML kigkonsult.se iCalcreator 2.20.2//
METHOD:PUBLISH
X-WR-CALNAME;VALUE=TEXT:Eventi DIAG
BEGIN:VTIMEZONE
TZID:Europe/Paris
BEGIN:STANDARD
DTSTART:20161030T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20170326T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.7458.field_data.0@www.ugov-ricerca.uniroma1.it
DTSTAMP:20260407T072433Z
CREATED:20161214T110818Z
DESCRIPTION:In this presentation I focus on two lines of research which sha
 re the aim of reusing visual knowledge. First\, I will discuss zero-exampl
 e classification.While there currently exists big (annotated) image and vi
 deo datasets\, these cannot guarantee sufficient annotations for all possi
 ble concepts. When considering more exotic concepts (e.g. lagerphone) or c
 omposite concepts (e.g.\, wooden saxophone\; a sunny day on the mountain)\
 , annotations are harder to obtain\, possibly can only obtained by experts
 \, and the number of (combinations of) concepts is (almost) unbounded. In 
 the absense of object specific annotations one solution is zero-shot learn
 ing\, where the combination of a) existing classifiers and b) semantic\, c
 ross-concept mappings between these classifiers allows for building novel 
 classifiers without expecting any visual examples. In particular I will fo
 cus on zero-shot learning for video retrieval and as prior for Second\, I 
 will discuss distance based classification with metric-learning\, where th
 e learned metric encapsulates our visual knowledge. The advantage of metri
 c learning is that it is trivial to add new classes or new concepts or new
  train data. In particular I will focus on metric learning for online clas
 sification of data streams.     
DTSTART;TZID=Europe/Paris:20161219T143000
DTEND;TZID=Europe/Paris:20161219T143000
LAST-MODIFIED:20190805T155749Z
LOCATION:B101
SUMMARY:Learning to Reuse Visual Knowledge - Thomas Mensink
URL;TYPE=URI:http://www.ugov-ricerca.uniroma1.it/node/7458
END:VEVENT
END:VCALENDAR