Algorithms for Approximate Minimization of the Difference Between Submodular Functions, with Applications

Abstract

We extend the work of Narasimhan and Bilmes [30] for minimizing set functions representable as a difference between submodular functions. Similar to [30], our new algorithms are guaranteed to monotonically reduce the objective function at every step. We empirically and theoretically show that the per-iteration cost of our algorithms is much less than [30], and our algorithms can be used to efficiently minimize a difference between submodular functions under various combinatorial constraints, a problem not previously addressed. We provide computational bounds and a hardness result on the mul- tiplicative inapproximability of minimizing the difference between submodular functions. We show, however, that it is possible to give worst-case additive bounds by providing a polynomial time computable lower-bound on the minima. Finally we show how a number of machine learning problems can be modeled as minimizing the difference between submodular functions. We experimentally show the validity of our algorithms by testing them on the problem of feature selection with submodular cost features.

17 pages, 8 figures. A shorter version of this appeared in Proc. Uncertainty in Artificial Intelligence (UAI), Catalina Islands, 2012

References in corpus (4)

• Near-optimal Nonmyopic Value of Information in Graphical Models
• Non-monotone submodular maximization under matroid and knapsack constraints
• Submodular approximation: sampling-based algorithms and lower bounds
• Maximizing General Set Functions by Submodular Decomposition

Cited by in corpus (13)

• Guarantees for Greedy Maximization of Non-submodular Functions with Applications
• Submodularity in Action: From Machine Learning to Signal Processing Applications
• Near-Optimal Sparse Sensing for Gaussian Detection with Correlated Observations
• Co-Mixup: Saliency Guided Joint Mixup with Supermodular Diversity
• Learning Interpretable Decision Rule Sets: A Submodular Optimization Approach
• Approximate Submodular Functions and Performance Guarantees
• Non-submodular Function Maximization subject to a Matroid Constraint, with Applications
• StratLearner: Learning a Strategy for Misinformation Prevention in Social Networks
• Opinion-aware Influence Maximization in Online Social Networks
• On the Approximation Relationship between Optimizing Ratio of Submodular (RS) and Difference of Submodular (DS) Functions
• Training Data Subset Selection for Regression with Controlled Generalization Error
• Majorisation-minimisation algorithms for minimising the difference between lattice submodular functions
• Regression Under Human Assistance