Robust and Interactive Information Gathering in Dynamic Environments
Abstract
In this proposal, we lay out a basic research program for tackling one of the fundamental challenges of our time: dealing with the overload of information. We seek to substantially advance information gathering from massive data streams and develop groundbreaking novel interactive methods to make effective decisions under uncertainty. We will demonstrate the performance of our methods on real-world problems that span a variety of disciplines. The proposed research is expected to lead to novel algorithms which will be directly applicable to the goals of the DoD and the ONR, such as intelligence gathering, surveillance, intrusion detection, and environmental monitoring. It will address the concern of massive data overload expressed by many military personnel [156] and help incorporate data science into military actions [157]. For example, the Navy recently announced a 200 million USD contract to purchase up to 141,500 sonobuoys for naval operations [158]. The problem of strategically placing these sensors and processing the information generated by them is a perfect fit for both the streaming and adaptive submodular maximization methods presented in this proposal. In addition, by only selecting the most informative data, our methods would reduce the computational and memory strains placed on the Distributed Common Ground System [159] used by the Navy, allowing for significantly faster performance. Our methods may also be applied to further increase the efficiency and ffectiveness of current condition-based maintenance (CBM) programs [160].
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 13, 2019
- Source ID
- N000141912406
Entities
People
- Amin Karbasi
Organizations
- Office of Naval Research
- United States Navy
- Yale University