We first modify the algorithm of Chekuri et al. (2010) to achieve a(1 1/e) approximation for m=o( klog3k),with 0ask .

In this work, we undertake a variational inference approach and approximate these rich distributions with simpler ones that respect the combinatorial constraints but are fully tractable. These approximations posses very strong negativeassociation properties, which we utilize inour theory.

Compared to robust submodular maximization for set function, new challenges arise when sequence functions are concerned.

Recently,there appear studies aiming atdesigning more efficient and practical algorithms for this problem.

Finally, we empirically demonstrate the effectiveness of our approach on several instances.