Data-Driven Online and Real-Time Combinatorial Optimization.

摘要

The main focus of our research has been on the fundamental aspects of optimization in the context of uncertain (and possibly large) data sets revealed in an online fashion, considering the intersection and interplay of three main phenomena (incomplete and uncertain data, online decisions with or without real-time restrictions, and large data sets). Motivated by applications associated with the deployment of autonomous multi-agent systems for spatial exploration and information harvesting, our research has concentrated on the development and analysis of competitive online algorithms for the simplest canonical models defined in our proposal (single agent prize collecting online traveling salesman problem and Hamiltonian path problems), as well as for some generalizations of the secretary problem, a class of closely related online problems.