Gradient Methods for Submodular Maximization

In this paper, we study the problem of maximizing continuous submodular functions that naturally arise in many learning applications such as those involving utility functions in active learning and sensing, matrix approximations and network inference. Despite the apparent lack of convexity in such functions, we prove that stochastic projected gradient methods can provide strong approximation guarantees for maximizing continuous submodular functions with convex constraints. More specifically, we prove that for monotone continuous DR-submodular functions, all fixed points of projected gradient ascent provide a factor $1/2$ approximation to the global maxima. We also study stochastic gradient and mirror methods and show that after $\mathcal{O}(1/\epsilon^2)$ iterations these methods reach solutions which achieve in expectation objective values exceeding $(\frac{\text{OPT}}{2}-\epsilon)$. An immediate application of our results is to maximize submodular functions that are defined stochastically, i.e. the submodular function is defined as an expectation over a family of submodular functions with an unknown distribution. We will show how stochastic gradient methods are naturally well-suited for this setting, leading to a factor $1/2$ approximation when the function is monotone. In particular, it allows us to approximately maximize discrete, monotone submodular optimization problems via projected gradient descent on a continuous relaxation, directly connecting the discrete and continuous domains. Finally, experiments on real data demonstrate that our projected gradient methods consistently achieve the best utility compared to other continuous baselines while remaining competitive in terms of computational effort.

[1]  Hui Lin,et al.  A Class of Submodular Functions for Document Summarization , 2011, ACL.

[2]  László Lovász,et al.  Submodular functions and convexity , 1982, ISMP.

[3]  Yurii Nesterov,et al.  Introductory Lectures on Convex Optimization - A Basic Course , 2014, Applied Optimization.

[4]  Yuichi Yoshida,et al.  A Generalization of Submodular Cover via the Diminishing Return Property on the Integer Lattice , 2015, NIPS.

[5]  Jan Vondrák,et al.  Maximizing a Submodular Set Function Subject to a Matroid Constraint (Extended Abstract) , 2007, IPCO.

[6]  P. Brucker Review of recent development: An O( n) algorithm for quadratic knapsack problems , 1984 .

[7]  Jure Leskovec,et al.  Inferring networks of diffusion and influence , 2010, KDD.

[8]  Andreas Krause,et al.  Near-optimal sensor placements in Gaussian processes , 2005, ICML.

[9]  Chandra Chekuri,et al.  Submodular function maximization via the multilinear relaxation and contention resolution schemes , 2011, STOC '11.

[10]  Andreas Krause,et al.  Guaranteed Non-convex Optimization: Submodular Maximization over Continuous Domains , 2016, AISTATS.

[11]  Sébastien Bubeck,et al.  Convex Optimization: Algorithms and Complexity , 2014, Found. Trends Mach. Learn..

[12]  Éva Tardos,et al.  Maximizing the Spread of Influence through a Social Network , 2015, Theory Comput..

[13]  Oluwasanmi Koyejo,et al.  Examples are not enough, learn to criticize! Criticism for Interpretability , 2016, NIPS.

[14]  Britta Peis,et al.  Submodular Function Maximization on the Bounded Integer Lattice , 2015, WAOA.

[15]  Andreas Krause,et al.  From MAP to Marginals: Variational Inference in Bayesian Submodular Models , 2014, NIPS.

[16]  Abhimanyu Das,et al.  Submodular meets Spectral: Greedy Algorithms for Subset Selection, Sparse Approximation and Dictionary Selection , 2011, ICML.

[17]  Ken-ichi Kawarabayashi,et al.  Optimal Budget Allocation: Theoretical Guarantee and Efficient Algorithm , 2014, ICML.

[18]  Benjamin Recht,et al.  Sharp Time–Data Tradeoffs for Linear Inverse Problems , 2015, IEEE Transactions on Information Theory.

[19]  Maryam Fazel,et al.  Designing smoothing functions for improved worst-case competitive ratio in online optimization , 2016, NIPS.

[20]  Mahdi Soltanolkotabi,et al.  Structured Signal Recovery From Quadratic Measurements: Breaking Sample Complexity Barriers via Nonconvex Optimization , 2017, IEEE Transactions on Information Theory.

[21]  Dafna Shahaf,et al.  Turning down the noise in the blogosphere , 2009, KDD.

[22]  Andreas Krause,et al.  Near-Optimally Teaching the Crowd to Classify , 2014, ICML.

[23]  Rishabh K. Iyer,et al.  Submodular Point Processes with Applications to Machine learning , 2015, AISTATS.

[24]  Andreas Krause,et al.  Stochastic Submodular Maximization: The Case of Coverage Functions , 2017, NIPS.

[25]  Andreas Krause,et al.  Cost-effective outbreak detection in networks , 2007, KDD '07.

[26]  藤重 悟 Submodular functions and optimization , 1991 .

[27]  Huy L. Nguyen,et al.  A Reduction for Optimizing Lattice Submodular Functions with Diminishing Returns , 2016, ArXiv.

[28]  Morteza Zadimoghaddam,et al.  Probabilistic Submodular Maximization in Sub-Linear Time , 2017, ICML.

[29]  Mahdi Soltanolkotabi,et al.  Learning ReLUs via Gradient Descent , 2017, NIPS.

[30]  Francis Bach,et al.  Submodular functions: from discrete to continuous domains , 2015, Mathematical Programming.

[31]  Yin Tat Lee,et al.  Subquadratic submodular function minimization , 2016, STOC.

[32]  Baharan Mirzasoleiman,et al.  Fast Constrained Submodular Maximization: Personalized Data Summarization , 2016, ICML.

[33]  Jack Edmonds,et al.  Matroids and the greedy algorithm , 1971, Math. Program..

[34]  Panos M. Pardalos,et al.  An algorithm for a singly constrained class of quadratic programs subject to upper and lower bounds , 1990, Math. Program..

[35]  Laurence A. Wolsey,et al.  An analysis of the greedy algorithm for the submodular set covering problem , 1982, Comb..

[36]  Jan Vondrák,et al.  On Multiplicative Weight Updates for Concave and Submodular Function Maximization , 2015, ITCS.

[37]  Andreas Krause,et al.  Distributed Submodular Maximization: Identifying Representative Elements in Massive Data , 2013, NIPS.

[38]  Rishabh K. Iyer,et al.  Submodular Optimization with Submodular Cover and Submodular Knapsack Constraints , 2013, NIPS.