2020
Deterministic Approximation for Submodular Maximization over a Matroid in Nearly Linear Time
NeurIPS 2020poster
We study the problem of maximizing a non-monotone, non-negative submodular function subject to a matroid constraint. The prior best-known deterministic approximation ratio for this problem is $\frac{1}{4}-\epsilon$ under $\mathcal{O}(({n^4}/{\epsilon})\log n)$ time complexity. We show that this dete…