BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Memento EPFL//
BEGIN:VEVENT
SUMMARY:Seminar by Prof. Immanuel Bomze\, University of Vienna
DTSTART:20190204T150000
DTEND:20190204T163000
DTSTAMP:20260916T225611Z
UID:a2d66f3d67a4ba6cece553b139be20c252e07958806cde2a9c820da2
CATEGORIES:Conferences - Seminars
DESCRIPTION:Prof. Immanuel Bomze\, University of Vienna\n\n"Robust cluster
 ing in social networks"\n\nAbstract\n\nDuring the last decades the importa
 nce of considering data uncertainty in optimization problems has become in
 creasingly apparent\, since small fluctuations of input data may lead to c
 omparably bad decisions in many practical problems when uncertainty is ign
 ored. If the probability distribution of the uncertain data is not known (
 or cannot be estimated with sufficient accuracy)\, a common technique is 
 to estimate bounds on the uncertain data (i.e.\, define uncertainty sets) 
 and to identify optimal solutions that are robust against data fluctuatio
 ns within these bounds. This approach leads to the robust optimization pa
 radigm that allows to consider uncertain objectives and constraints. Optim
 ization problems where only the objective is uncertain arise\, for instan
 ce\, prominently in the analysis of social networks. This stems from the 
 fact that the strength of social ties (i.e.\, the amount of influence ind
 ividuals exert on each other) or the willingness of individuals to adopt 
 and share information can\, for example\, only be roughly estimated based
  on observations. A fundamental problem arising in social network analysi
 s regards the identification of communities (e.g.\, work groups\, interes
 t groups)\, which can be modeled as a Dominant Set Clustering Problem whi
 ch in turn leads to a Standard Quadratic Optimization Problems (StQP). He
 re the link strengths enter the objective while the constraints are famil
 iar probability constraints\, so that they can be considered certain. Hen
 ce we investigate data uncertainty in the objective function of StQPs\, c
 onsidering different uncertainty sets\, and derive implications for the c
 omplexity of robust variants of the corresponding deterministic counterpa
 rts. We can show that considering data uncertainty in a StQP results in a
 nother StQP of the same complexity if ellipsoidal\, spherical or boxed un
 certainty sets are assumed. Moreover we discuss implications when conside
 ring polyhedral uncertainty sets\, and derive rigorous bounds for this ca
 se.
LOCATION:ODY 4 03 https://plan.epfl.ch/?room==ODY%204%2003
STATUS:CONFIRMED
END:VEVENT
END:VCALENDAR