Resilient Non-Submodular Maximization over Matroid Constraints
share
Applications in control, robotics, and optimization motivate the design of systems by selecting system elements, such as actuators, sensors, or data, subject to complex design constraints that require the system elements not only to be a few in number, but also, to satisfy heterogeneity or global-interdependency constraints; in particular, matroid constraints. However, in failure-prone and adversarial environments, sensors get attacked; actuators fail; data get deleted. Thence, traditional matroid-constrained design paradigms become insufficient and, in contrast, resilient matroid-constrained designs against attacks, failures, or deletions become important. In general, resilient matroid-constrained design problems are computationally hard. Also, even though they often involve objective functions that are monotone and (possibly) submodular, no scalable approximation algorithms are known for their solution. In this paper, we provide the first algorithm, that achieves the following characteristics: system-wide resiliency, i.e., the algorithm is valid for any number of denial-of-service attacks, deletions, or failures; minimal running time, i.e., the algorithm terminates with the same running time as state-of-the-art algorithms for (non-resilient) matroid-constrained optimization; and provable approximation performance, i.e., the algorithm guarantees for monotone objective functions a solution close to the optimal. We quantify the algorithm's approximation performance using a notion of curvature for monotone (not necessarily submodular) set functions. Finally, we support our theoretical analyses with numerical experiments, by considering a control-aware sensor selection scenario, namely, sensing-constrained robot navigation.
READ FULL TEXT
