BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//wp-events-plugin.com//7.2.3.1//EN
TZID:Europe/Paris
X-WR-TIMEZONE:Europe/Paris
BEGIN:VEVENT
UID:1137@cran.univ-lorraine.fr
DTSTART;TZID=Europe/Paris:20251010T140000
DTEND;TZID=Europe/Paris:20251010T150000
DTSTAMP:20251009T145739Z
URL:https://www.cran.univ-lorraine.fr/events/seminaire-de-pavlo-mozharovsk
 yi/
SUMMARY:Séminaire de Pavlo Mozharovskyi
DESCRIPTION:Speaker: Pavlo Mozharovskyi (Télécom Paris)\nTitle: Explainab
 le anomaly detection using data depth\nLocation: seminar room\, Henri Poin
 care building\, 4th floor\n\nAbstract: Anomaly detection is a branch of da
 ta analysis and machine learning which aims at identifying observations th
 at exhibit abnormal behaviour. Be it measurement errors\, disease developm
 ent\, severe weather\, production quality default(s) (items) or failed equ
 ipment\, financial frauds or crisis events\, their on-time identification\
 , isolation and explanation constitute an important task in almost any bra
 nch of science and industry. By providing a robust ordering\, data depth -
  statistical function that measures belongingness of any point of the spac
 e to a data set - becomes a particularly useful tool for detection of anom
 alies. Already known for its theoretical properties\, data depth has under
 gone substantial computational developments in the last decade and particu
 larly recent years\, which has made it applicable for contemporary-sized p
 roblems of data analysis and machine learning.\n\nWe study data depth as a
 n efficient anomaly detection tool\, assigning abnormality labels to obser
 vations with lower depth values\, in a multivariate setting. Practical que
 stions of necessity and reasonability of invariances and shape of the dept
 h function\, their robustness and computational complexity\, choice of the
  threshold are discussed. Furthermore\, we introduce a new statistical too
 l dedicated for exploratory analysis of abnormal observations using data d
 epth as a score. Abnormal component analysis (shortly ACA) is a method tha
 t searches a low-dimensional data representation which best visualises and
  explains anomalies. This low-dimensional representation not only allows t
 o distinguish groups of anomalies better than the methods of the state of 
 the art\, but as well provides a -- linear in variables and thus easily in
 terpretable -- explanation for anomalies. Illustrations include use-cases 
 that underline advantageous behaviour of data depth and of the explainable
  anomaly detection\, in various settings.\n
CATEGORIES:Séminaires projet SiMul
LOCATION:CRAN - FST - 4ème\, Campus Sciences\, Boulevard des Aiguillettes\
 , Vandoeuvre-lès-Nancy\, 54506\, France
X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=Campus Sciences\, Boulevard
  des Aiguillettes\, Vandoeuvre-lès-Nancy\, 54506\, France;X-APPLE-RADIUS=
 100;X-TITLE=CRAN - FST - 4ème:geo:0,0
END:VEVENT
BEGIN:VTIMEZONE
TZID:Europe/Paris
X-LIC-LOCATION:Europe/Paris
BEGIN:DAYLIGHT
DTSTART:20250330T030000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
END:VCALENDAR