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:20131027T030000
TZOFFSETFROM:+0200
TZOFFSETTO:+0100
TZNAME:CET
END:STANDARD
BEGIN:DAYLIGHT
DTSTART:20130331T020000
TZOFFSETFROM:+0100
TZOFFSETTO:+0200
TZNAME:CEST
END:DAYLIGHT
END:VTIMEZONE
BEGIN:VEVENT
UID:calendar.6672.field_data.0@www.ugov-ricerca.uniroma1.it
DTSTAMP:20260407T214008Z
CREATED:20130606T122513Z
DESCRIPTION:Prof. A. Fazel Famili\, Principal Research Scientist of the Nat
 ional Research Council (NRC) of Canada and Professor at the University of 
 Ottawa\, will give the following seminary during his scientific visit to o
 ur  Department:'Analyzing Imbalanced Data in Life Sciences: Methods and Al
 ternatives'Over the last 10-20 years\, extensive developments of tools and
  software systems to design experiments\, automatically monitor\, collect 
 and warehouse large amounts of data\, from applications such as life scien
 ces and industrial processes\, have been the prime motivation for an evolv
 ing data mining paradigm. This has created several challenges among which 
 are situations where the data is imbalanced. The class imbalance problem c
 orresponds to cases where majority of samples belong to one class and a sm
 all minority belongs to the other\, which in many cases is equally or even
  more important. The goal is still to learn from this data and discover mo
 dels that are useful and unknown to the producers of this data. Most machi
 ne learning algorithms are overwhelmed by the majority class and ignore th
 e minority class since the traditional classifiers focus more on minimizin
 g the overall error rate instead of paying special attention to the minori
 ty class. This could result in classifying all the data into the majority 
 class in order to achieve higher accuracy. As an example\, decision trees 
 tend to over-generalize the class that is represented by most of the examp
 les in the data. This obviously creates a major problem.To deal with this 
 problem a number of approaches have been studied in the past and have been
  applied to many domains. In this talk we will provide an overview of some
  existing methods. We will then introduce a novel approach that is differe
 nt than almost all existing ones and is based on identifying the inherent 
 characteristics of one class vs the other. We present the results of our s
 tudies focusing on real data from life science applications.
DTSTART;TZID=Europe/Paris:20130618T120000
DTEND;TZID=Europe/Paris:20130618T120000
LAST-MODIFIED:20130617T130553Z
LOCATION:Aula Magna del DIAG\, Via Ariosto 25
SUMMARY:MORE@DIAG Seminary: Analyzing Imbalanced Data in Life Sciences - Pr
 of. Fazel Famili\, National Research Council (NRC) of Canada and Universit
 y of Ottawa
URL;TYPE=URI:http://www.ugov-ricerca.uniroma1.it/node/6672
END:VEVENT
END:VCALENDAR