Although submodular maximization generalizes many fundamental problems in discrete optimization, lots of real-world problems are non-submodular. In this paper, we consider the maximization problem of non-submodular function with a knapsack constraint, and explore the performance of the greedy algorithm. Our guarantee is characterized by the submodularity ratio β and curvature α. In particular, we prove that the greedy algorithm enjoys a tight approximation guarantee of (Formula Presented) for the above problem. To our knowledge, it is the first tight constant factor for this problem. In addition, we experimentally validate our algorithm by an important application, the Bayesian A-optimality.
CITATION STYLE
Zhang, Z., Liu, B., Wang, Y., Xu, D., & Zhang, D. (2019). Greedy Algorithm for Maximization of Non-submodular Functions Subject to Knapsack Constraint. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 11653 LNCS, pp. 651–662). Springer Verlag. https://doi.org/10.1007/978-3-030-26176-4_54
Mendeley helps you to discover research relevant for your work.