(YIP) DATA ACQUISITION IN DYNAMIC ENVIRONMENTS: A SUBMODULAR PERSPECTIVE
Abstract
Whether we select a bunch of sensory observations, or choose a sequence of actions, or collaborate with a number of agents, the data-acquisition task often involves inherent combinatorial structures and is fundamentally discrete. Even though discrete optimization problems are generally hard, prior work has shown that many data-acquisition problems admit a key structural property called submodularity. Due to recent breakthroughs in exploiting submodularity for discrete optimization, we now have efficient and provable algorithms for special cases of the data-acquisition problem. However, designing discrete and submodular optimization methodologies that are capable of adapting to dynamic and uncertain environments requires a quantum leap in the following three main directions, as aimed by this proposal: (i) developing foundational tools for discrete optimization in complex environments addressing uncertainty, resiliency, and unknown dynamics; (ii) designing polices that sequentially select the most informative data and observations while learning and adapting to the environment; (iii) developing cooperative strategies among multiple agents to jointly achieve similar goals as in (i),(ii). We are also committed to evaluating the performance of our methods on real world problems relevant to the US Air Force.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Aug 12, 2021
- Source ID
- FA95502010111
Entities
People
- Hamed Hassani
Organizations
- Air Force Office of Scientific Research
- United States Air Force
- University of Pennsylvania