09h00 - 10h00
Data Driven Online Resource Allocation Problems
In this talk we are interested in a general class of data driven online resource allocation problems exhibiting a combination of (i) incomplete and uncertain input streams revealed over time, (ii) time-sensitive objectives, and (iii) computational constraints for making online decisions. After discussing some contexts and applications for this class of problems (sponsored search auctions and online auctions; on-demand video/movie requests; yield management; kidney exchanges) we will discuss in details various methodological and algorithmic results we have obtained on some basic canonical problems from this class: online bipartite matching problems, online matroidal secretary problems, and online linear programming. Research funded in part by NSF, ONR, and AFOSR.