Nos próximos dias 27 de Novembro e 4 de Dezembro, entre as 9h30 e as 11h, na sala S3 (FC6-1.42) do DCC, Marco Silva irá dar um tutorial sobre o tema ""Optimization under uncertainty".
O tutorial é organizado pelo DCC-FCUP e pelo grupo de investigação CEGI-INESCTEC e é aberto a todos os interessados.
Short Bio
Marco Silva is a post-doctoral researcher at CEGI-INESCTEC. His research during the last years has focused on optimization under data uncertainty. Marco has conducted this research
under a PhD "cotutelle" (dual degree) program between University of Avignon, France, and Federal University of Rio de Janeiro, Brazil. He has been working in the intersection of the disciplines of robust optimization, stochastic optimization, machine learning and integer programming.
His work leverages the existing knowledge under these disciplines in order to reformulate uncertain problems where the associated probability distributions are unknown. He has been interested in resolution methods (exact and heuristic approximations) for the related type of problems.
Marco has a past international executive career within the information systems and infrastructure management industry. He uses this experience in order to link research to practical applications under a systemic approach.
"Optimization under uncertainty: a tutorial"
Abstract
Within a two section presentation we will review relevant modeling approaches to optimization under uncertainty, when the problem data cannot be known accurately. A key difficulty in optimization under uncertainty is in dealing with an uncertainty space that
is huge and frequently leads to very large-scale optimization models.
Decision-making under uncertainty is often further complicated by the presence of integer decision variables to model logical constraints in a multi-stage settings. Under this overview perspective, we discuss the classical stochastic programming, robust programming
and distributionally robust programming linear approaches. The advantages and shortcomings of these models are reviewed and a special focus is given to possible tractable reformulations of distributionally robust problems.