BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20260512T045336EDT-6424wCH4i1@132.216.98.100
DTSTAMP:20260512T085336Z
DESCRIPTION:Enabling model-driven analytics for cyber physical systems.\n\n
 During this talk\, I will present what we call model-driven live analytics
 . I will focus on the key enablers to make this approach scalable to the s
 ize of country-wide CPSs. The main contribution of this work is a multi-di
 mensional graph data model that brings raw data\, domain knowledge\, and m
 achine learning together in a single model\, which can drive live analytic
  processes. Firstly\, data handled by cyber-physical systems is usually dy
 namic and changes frequently and at different paces. I will present a temp
 oral graph data model and storage system\, which consider time as a first-
 class property and allow to analyse frequently changing data. Additionally
 \, I will present how a continuous sequence of sensor values can be effici
 ently encoded using live mathematical model inference. Secondly\, making s
 ustainable decisions requires to anticipate which impacts certain actions 
 could have. In some cases\, hundreds or thousands of such hypothetical act
 ions must be explored ahead before any solid decision can be taken. I will
  present our approach to deal with such need - a multi-dimensional model t
 hat efficiently represent\, store and analyse many different alternatives 
 of the same system in live. Thirdly\, to make smart decisions\, cyber-phys
 ical systems must continuously refine behavioural models that are known at
  design time\, with what can only be learned from live data. During this t
 alk\, I will present how we have combined machine learning and the multi-d
 imensional graph data model to empower live analytics for cyber-physical s
 ystems. Finally\, I will conclude this talk with details about the open so
 urce project\, which is developed around these research ideas\, and share 
 the lessons learned with respect to high-performance Java and big data.\n	
 \n	-----> Cette présentation sera donnée en français. \n
DTSTART:20170130T203000Z
DTEND:20170130T213000Z
LOCATION:Room 3195\, CA\, QC\, Montreal\, H3T 1J4\, Pavillon André-Aisensta
 dt\, 2920\, Chemin de la tour\, 5th floor
SUMMARY:François Fouquet\, University of Luxembourg
URL:/mathstat/channels/event/francois-fouquet-universi
 ty-luxembourg-265448
END:VEVENT
END:VCALENDAR