Abstract
We study an extension of the classical graph cut problem, wherein we replace the modular (sum of edge weights) cost function by a submodular set function defined over graph edges. Special cases of this problem have appeared in different applications in signal processing, machine learning, and computer vision. In this paper, we connect these applications via the generic formulation of “cooperative graph cuts”, for which we study complexity, algorithms, and connections to polymatroidal network flows. Finally, we compare the proposed algorithms empirically.
Similar content being viewed by others
Notes
Undirected graphs can be reduced to bidirected graphs.
We will also call it the ellipsoidal approximation since it is based on approximating a symmetrized version of the submodular polyhedron by an ellipsoid.
The expansion is described in Section 10.3 in Narayanan [57]. In short, we replace each element e by a set \(\hat{e}\) of f(e) parallel elements. Thereby we extend f to a submodular function \(\hat{f}\) on subsets of \(\bigcup _i \hat{e}_i\). The desired rank function is now the convolution \(r(\cdot ) = \hat{f}(\cdot ) *|\cdot |\) and it satisfies \(f(S) = r(\bigcup _{e \in S}\hat{e})\).
A cut \(C \subseteq \mathcal {E}\) is minimal if no proper subset \(B \subset C\) is a cut.
If \( \min _{e \in P_{\min }}f(e|C)=0\), then we greedily pick all such edges with zero marginal cost, because they do not increase the cost. Otherwise we sample as indicated in the algorithm.
For unconstrained submodular function minimization we drop the constraint that the functions \(g_j\) are nondecreasing.
These are distinct from the other previously-used notion of symmetric submodular functions [59] where, for all \(A \subseteq \mathcal {E}\), \(f(A) = f(\mathcal {E}{\setminus } A)\).
An undirected graph can easily be turned into a directed one by replacing each edge by two opposing directed ones that have the same cost. A cut will always only include one of those edges.
Most of the bounds proved above are absolute, and not asymptotic. The only exception is \(\hat{f}_{ea}\). For simplicity, it is here treated as an absolute bound.
References
Ahuja, R.K., Magnanti, T.L., Orlin, J.B.: Network Flows. Prentice Hall, Upper Saddle River (1993)
Allène, C., Audibert, J.-Y., Couprie, M., Keriven, R.: Some links between extremum spanning forests, watersheds, and min-cuts. Image Vision Comput. 28(10), 1460–1471 (2009)
Bagon, S.: Matlab wrapper for graph cut. http://www.wisdom.weizmann.ac.il/~bagon (December 2006)
Balcan, N., Harvey, N.: Submodular functions: learnability, structure, and optimization. arXiv:1008.2159v3 [cs.DS] (2012)
Baumann, F., Berckey, S., Buchheim, C.: Facets of Combinatorial Optimization—Festschrift for Martin Grötschel, Chapter Exact Algorithms for Combinatorial Optimization Problems with Submodular Objective Functions, pp. 271–294. Springer, New York (2013)
Bilmes, J.: Dynamic graphical models—an overview. IEEE Signal Process. Mag. 27(6), 29–42 (2010)
Bilmes, J., Bartels, C.: On triangulating dynamic graphical models. In: Uncertainty in Artificial Intelligence, pp. 47–56. Morgan Kaufmann Publishers, Acapulco, Mexico (2003)
Boykov, Y., Jolly, M.-P.: Interactive graph cuts for optimal boundary and region segmentation of objects in n-d images. In: International Conference on Computer Vision (ICCV) (2001)
Boykov, Y., Kolmogorov, V.: An experimental comparison of min-cut/max-flow algorithms for energy minimization in vision. IEEE Trans. Pattern Anal. Mach. Intell. 26(9), 1124–1137 (2004)
Boykov, Y., Veksler, O.: Handbook of Mathematical Models in Computer Vision, Chapter Graph Cuts in Vision and Graphics: Theories and Applications. Springer, New York (2006)
Bunke, F., Hamacher, H.W., Maffioli, F., Schwahn, A.: Minimum cut bases in undirected networks. Report in Wirtschaftsmathematik (WIMA Report) 108, Universität Kaiserslautern (2007)
Chambolle, A., Darbon, J.: On total variation minimization and surface evolution using parametric maximum flows. Int. J. Comput. Vis. 84(3), 288–307 (2009)
Chekuri, C., Kannan, S., Raja, A., Viswanath, P.: Multicommodity flows and cuts in polymatroidal networks. In: Innovations in Theoretical Computer Science (ITCS) (2012)
Conforti, M., Cornuéjols, G.: Submodular set functions, matroids and the greedy algorithm: tight worst-case bounds and some generalizations of the Rado-Edmonds theorem. Discret. Appl. Math. 7(3), 251–274 (1984)
Couprie, C., Grady, L., Talbot, H., Najman, L.: Combinatorial continuous maximum flow. SIAM J. Imag. 4(3), 905–930 (2011)
Cunningham, W.H.: Decomposition of submodular functions. Combinatorica 3(1), 53–68 (1983)
Dantzig, G., Fulkerson, D.: On the max flow min cut theorem of networks. Technical Report P-826, The RAND Corporation (1955)
Ene, A., Vondrák, J., Wu, Y.: Local distribution and the symmetry gap: approximability of multiway partitioning problems. In: Proceedings of SIAM-ACM Symposium on Discrete Algorithms (SODA) (2013)
Fix, A., Joachims, T., Park, S.M., Zabih, R.: Structured learning of sum-of-submodular higher order energy functions. In: International Conference on Computer Vision (ICCV) (2013)
Ford, L., Fulkerson, D.: Maximal flow through a network. Can. J. Math. 8, 399–404 (1956)
Garey, M.R., Johnson, D.S., Stockmeyer, L.: Some simplified NP-complete graph problems. Theoret. Comput. Sci. 1(3), 237–267 (1976)
Goel, G., Karande, C., Tripati, P., Wang, L.: Approximability of combinatorial problems with multi-agent submodular cost functions. In: Proceedings of IEEE Symposium on Foundations of Computer Science (FOCS) (2009)
Goel, G., Tripathi, P., Wang, L.: Combinatorial problems with discounted price functions in multi-agent systems. In: Foundations of Software Technology and Theoretical Computer Science (FSTTCS) (2010)
Goemans, M., Harvey, N.J.A., Iwata, S., Mirrokni, V.S.: Approximating submodular functions everywhere. In: Proceedings of SIAM-ACM Symposium on Discrete Algorithms (SODA) (2009)
Goemans, M.X., Harvey, N.J.A., Kleinberg, R., Mirrokni, V.S.: On learning submodular functions—a preliminary draft. Unpublished Manuscript
Goyal, V., Ravi, R.: An FPTAS for minimizing a class of low-rank quasi-concave functions over a convex domain. Technical Report 366, Tepper School of Business, Carnegie Mellon University (2008)
Greig, D.M., Porteous, B.T., Seheult, A.H.: Exact maximum a posteriori estimation for binary images. J. R. Stat. Soc. 51(2), 271–279 (1989)
Hassin, R.: Minimum cost flow with set constraints. Networks 12, 1–21 (1982)
Hassin, R., Monnot, J., Segev, D.: Approximation algorithms and hardness results for labeled connectivity problems. J. Comb. Optim. 14(4), 437–453 (2007)
Heng, L., Gotovos, A., Krause, A., Pollefeys, M.: Efficient visual exploration and coverage with a micro aerial vehicle in unknown environments. In: IEEE International Conference on Robotics and Automation (ICRA) (2015)
Iwata, S., Nagano, K.: Submodular function minimization under covering constraints. In: Proceedings of IEEE Symposium on Foundations of Computer Science (FOCS) (2009)
Iyer, R., Bilmes, J.: Submodular optimization with submodular cover and submodular knapsack constraints. In: Neural Information Processing Society (NIPS) (2013)
Iyer, R., Jegelka, S., Bilmes, J.: Fast semidifferential-based submodular function optimization. In: Proceedings of International Conference on Machine Learning (ICML) (2013a)
Iyer, R., Jegelka, S., Bilmes, J.: Curvature and optimal algorithms for learning and minimizing submodular functions. In: Neural Information Processing Society (NIPS) (2013b)
Jegelka, S.: Combinatorial Problems with submodular coupling in machine learning and computer vision. PhD thesis, ETH Zurich (2012)
Jegelka, S., Bilmes, J.: Submodularity beyond submodular energies: coupling edges in graph cuts. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2011a)
Jegelka, S., Bilmes, J.: Approximation bounds for inference using cooperative cuts. In: Proceedings of International Conference on Machine Learning (ICML) (2011b)
Jegelka, S., Lin, H., Bilmes, J.: On fast approximate submodular minimization. In: Neural Information Processing Society (NIPS) (2011)
Jegelka, S., Bach, F., Sra, S.: Reflection methods for user-friendly submodular optimization. In: Neural Information Processing Society (NIPS) (2013)
Jha, S., Sheyner, O., Wing, J.: Two formal analyses of attack graphs. In: Proceedings of the 15th Computer Security Foundations Workshop, pp. 49–63 (2002)
Kannan, S., Viswanath, P.: Multiple-unicast in fading wireless networks: a separation scheme is approximately optimal. In: IEEE International Symposium on Information Theory (ISIT) (2011)
Kannan, S., Raja, A., Viswanath, P.: Local phy + global flow: a layering principle for wireless networks. In: IEEE International Symposium on Information Theory (ISIT) (2011)
Khalil, E.B., Dilkina, B., Song, L.: Scalable diffusion-aware optimization of network topology. In: Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 1226–1235. ACM, New York (2014)
Kohli, P., Kumar, M., Torr, P.: \(\text{P}^3\) & beyond: move making algorithms for solving higher order functions. IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 1645–1656 (2009a)
Kohli, P., Ladický, L., Torr, P.: Robust higher order potentials for enforcing label consistency. Int. J. Comput. Vision 82(3), 302–324 (2009b)
Kohli, P., Osokin, A., Jegelka, S.: A principled deep random field for image segmentation. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2013)
Kolmogorov, V.: Minimizing a sum of submodular functions. Discret. Appl. Math. 160(15), 2246–2258 (2012)
Kolmogorov, V., Boykov, Y.: What metrics can be approximated by geo-cuts, or global optimization of length/area and flux. In: International Conference on Computer Vision (ICCV) (2005)
Kolmogorov, V., Zabih, R.: What energy functions can be minimized via graph cuts? IEEE Trans. Pattern Anal. Mach. Intell. 26(2), 147–159 (2004)
Koufogiannakis, C., Young, N.E.: Greedy \(\Delta \)-approximation algorithm for covering with arbitrary constraints and submodular costs. In: International Colloquium on Automata, Languages and Programming (ICALP) (2009)
Krause, A.: SFO: a toolbox for submodular function optimization. J. Mach. Learn. Res. 11, 1141–1144 (2010)
Lawler, E.L., Martel, C.U.: Computing maximal “Polymatroidal” network flows. Math. Oper. Res. 7(3), 334–347 (1982)
Lin, H., Bilmes, J.: Learning mixtures of submodular shells with application to document summarization. In: Uncertainty in Artificial Intelligence (UAI) (2012)
Lovász, L.: Mathematical Programming—The State of the Art, Chapter Submodular Functions and Convexity, pp. 235–257. Springer, New York (1983)
Mittal, S., Schulz, A.: An FPTAS for optimizing a class of low-rank functions over a polytope. Math. Program. (2012)
Mitzenmacher, M., Upfal, E.: Probability and Computing: Randomized Algorithms and Probabilistic Analysis. Cambridge University Press, Cambridge (2005)
Narayanan, H.: Submodular Functions and Electrical Networks. Elsevier Science, New York (1997)
Nikolova, E.: Approximation algorithms for reliable stochastic combinatorial optimization. In: APPROX (2010)
Queyranne, M.: Minimizing symmetric submodular functions. Math. Program. 82, 3–12 (1998)
Ramalingam, S., Kohli, P., Alahari, K., Torr, P.: Exact inference in multi-label CRFs with higher order cliques. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2008)
Ramalingam, S., Russell, C., Ladicky, L., Torr, P.H.S.: Efficient minimization of higher order submodular functions using monotonic boolean functions. arXiv:1109.2304 (2011)
Rudin, L.I., Osher, S., Fatemi, E.: Nonlinear total variation based noise removal algorithms. Phys. D 60, 259–268 (1992)
Schrijver, A.: Combinatorial Optimization. Springer, New York (2004)
Shelhamer, E., Jegelka, S., Darrell, T.: Communal cuts: sharing cuts across images. In: NIPS Workshop on Discrete Optimization in Machine Learning (2014)
Silberman, N., Shapira, L., Gal, R., Kohli, P.: A contour completion model for augmenting surface reconstructions. In: Europen Conference on Computer Vision (ECCV) (2014)
Sinop, A., Grady, L.: A seeded image segmentation framework unifying graph cuts and random walker which yields a new algorithm. In: International Conference on Computer Vision (ICCV) (2007)
Stanley, R.P.: Enumerative Combinatorics, Volume I of Cambridge Studies in Advanced Mathematics. Cambridge University Press, Cambridge (1997)
Stobbe, P., Krause, A.: Efficient minimization of decomposable submodular functions. In: Neural Information Processing Society (NIPS) (2010)
Stobbe, P., Krause, A.: Learning Fourier Sparse Set Functions (2012)
Svitkina, Z., Fleischer, L.: Submodular approximation: Sampling-based algorithms and lower bounds. In: Proceedings of IEEE Symposium on Foundations of Computer Science (FOCS) (2008)
Taniai, T., Matsushita, Y., Naemura, T.: Superdifferential cuts for binary energies. In: IEEE Conference on Computer Vision and Pattern Recognition (CVPR) (2015)
Tardos, E., Tovey, C.A., Trick, M.A.: Layered augmenting path algorithms. Math. Oper. Res. 11(2), 362–370 (1986)
Tschiatschek, S., Iyer, R., Wei, H., Bilmes, J.: Learning mixtures of submodular functions for image collection summarization. In: Neural Information Processing Society (NIPS) (2014)
Vondrák, J.: Submodularity and curvature: the optimal algorithm. RIMS Kôkyûroku Bessatsu (2008)
Vondrák, J.: Symmetry and approximability of submodular maximization problems. SIAM J. Comput. 42(1), 265–304 (2013)
Zhang, P., Cai, J.-Y., Tang, L.-Q., Zhao, W.-B.: Approximation and hardness results for label cut and related problems. J. Comb. Optim. 21(2), 192–208 (2011)
Z̆ivný, S., Cohen, D.A., Jeavons, P.G.: The expressive power of binary submodular functions. Discret. Appl. Math. 157(15), 3347–3358 (2009)
Acknowledgments
We would like to thank Bernhard Schölkopf for the productive time at the Max Planck Institute in Tübingen, and Jens Vygen and Raymond Hemmecke for discussions. Prof. Bilmes would like to acknowledge support by the National Science Foundation under Grant No. IIS-1162606, the National Institutes of Health under award R01GM103544, by a Google, a Microsoft, and an Intel research award, and also support by TerraSwarm, one of six centers of STARnet, a Semiconductor Research Corporation program sponsored by MARCO and DARPA.
Author information
Authors and Affiliations
Corresponding author
Electronic supplementary material
Below is the link to the electronic supplementary material.
Appendices
Appendix: Cooperative multi-cut and sparsest cut
An extension of MinCoopCut is the problem of cooperative multi-way cut and sparsest cut. Using the approximation \(\hat{f}_{ea}\) from Sect. 5.1.3, we can transform any multi-way or sparsest cut problem with a submodular cost function on edges (instead of a sum of edge weights) into a cut problem whose cut cost is a convolution of local submodular functions. The relaxation of this cut problem is dual to the polymatroidal flow problems considered by Chekuri et al. [13]. Combining their results with ours, we get the following Lemma.
Lemma 11
Let \(\alpha \) be the approximation factor for solving a sparsest cut / multi-way cut in a polymatroidal network. If we solve a cooperative sparsest cut / multi-way cut by first approximating the cut cost f by a function \(\hat{f}_{ea}\) and, on this instance, using the method with factor \(\alpha \), we get an \(O(\alpha n)\)-approximation for cooperative sparsest cut / multi-way cut.
Using Theorems 6 and 8 in Chekuri et al. [13], we obtain the following bounds:
Corollary 2
There is an \(O(n\log k)\) approximation for cooperative sparsest cut in undirected graphs that is dual to a maximum multicommodity flow problem with k pairs, and an \(O(n\log k)\) approximation for cooperative multi-way cut.
We leave it as an open problem whether these bounds are optimal.
Proof of proposition 2
The first part of Proposition 2 is proven by Fig. 1. Here, we show the second part that the function \(h(X) = f(\delta ^+(X))\) is subadditive if f is nondecreasing and submodular. Let \(X,Y \subseteq \mathcal {V}\). Then it holds that
In Inequality (46), we used that f is submodular, and in Inequality (47), we used that f is nonnegative.
Reduction from Graph Bisection to MinCoopCut
In this section, we prove Theorem 2 via a reduction from Graph Bisection, which is known to be NP-hard [21].
Definition 2
(Graph Bisection) Given a graph \(\mathcal {G}_B = (\mathcal {V}_B,\mathcal {E}_B)\) with edge weights \(w_B: \mathcal {E}_B \rightarrow {\mathbb {R}}_+\), find a partition \(V_1 \dot{\cup }V_2 = \mathcal {V}_B\) with \(|V_1| = |V_2| = |\mathcal {V}_B|/2\) with minimum cut weight \(w(\delta (V_1))\).
Proof
To reduce Graph Bisection to MinCoopCut, we construct an auxiliary graph \(\mathcal {G}= (\mathcal {V}_B \cup \{s,t\}, \mathcal {E}_B \cup \mathcal {E}_s \cup \mathcal {E}_t)\) that contains an unchanged copy of \(\mathcal {G}_B\) and two additional terminal nodes s, t. The submodular weights on the edges adjacent to the terminal nodes will express the balance constraint \(|V_1| = |V_2| = |\mathcal {V}_B|/2\). In \(\mathcal {G}\), we retain the modular costs w on \(\mathcal {E}_B\) and connect s, t to every vertex in \(\mathcal {G}_B\) with corresponding new edge sets \(\mathcal {E}_s\) and \(\mathcal {E}_t\), as illustrated in Fig. 7a. The cost of a cut in \(\mathcal {G}\) is measured by the submodular function
where \(\beta \) is an appropriately large constant, and \(f_{bal}\) will be defined later. Obviously, any minimal (s, t)-cut C must include \(n_B=|\mathcal {V}_B|\) edges from \(\mathcal {E}_s \cup \mathcal {E}_t\), and partitions \(\mathcal {V}_B\). Moreover, the cardinality of \(C_s = C \cap \mathcal {E}_s\) is the number of nodes in \(\mathcal {V}_B\) assigned to t. Hence, in an equipartition, \(|C_s| = |C_t| = n_B/2\), where \(C_t = C \cap \mathcal {E}_t\).
It remains to define \(f_{bal}\) as a nondecreasing submodular function that implements the equipartition constraint. The function will be an expectation over matroid rank functions \(h_\sigma \). Let \(\mathcal {H}_\sigma = (\mathcal {E}_s, \mathcal {E}_t, \mathcal {F}_{\sigma })\) be a bipartite graph with nodes \(\mathcal {E}_s \cup \mathcal {E}_t\) whose edges \(\mathcal {F}_{\sigma }\) form a derangement between \(\mathcal {E}_s\) and \(\mathcal {E}_t\), as illustrated in Fig. 7b.
Let \(\phi (C_s \cup C_t)\) be the image of \(C_s \cup C_t\) in the set of nodes of \(\mathcal {H}_\sigma \). The function \(h_{\sigma }: 2^{\phi (\mathcal {E}_s \cup \mathcal {E}_t)} \rightarrow {\mathbb {N}}_0\) counts the number of connected components in the subgraph induced by the nodes \(\phi (C_s \cup C_t)\), and is the rank of a partition matroid. Figure 7 shows some examples.
Let \({\mathfrak {S}}\) be the set of all derangements \(\sigma \) of \(n_B\) elements, i.e., all possible edge configurations in \(\mathcal {H}_\sigma \). We define \(f_{bal}\) to be the expectation (under uniform draws of \(\sigma \))
For a fixed derangement \(\sigma '\) and a fixed size \(|C_s \cup C_t|=n_B\), the value \(h_{\sigma '}(C_s \cup C_t)\) is minimal if \(\sigma '(C_s) = C_t\) and \(|C_s| = |C_t|\). For a fixed \(\sigma \), the rank \(h_{\sigma }(C)\) is \(|\phi (C_s \cup C_t)| = |C_s|+|C_t|\) minus the matching nodes. Denoting \((s,v_i)\) in \(\mathcal {H}_{\sigma }\) by \(x_i\) and \((v_i,t)\) by \(y_i\), the rank is
Hence, the sum in (51) becomes
To compute the sum over \(\sigma \) in the second term, let \(C_{s\cap t} \triangleq \{ (s, v)\, |\, \{(s,v),(v, t)\} \subseteq C\}\) be the set of s-edges whose counterpart on the t side is also contained in C. Let further \(D'(n_B-1)\) denote the number of permutations of \(n_B-1\) elements (pair \((x_i,y_k)\), i.e., \(\sigma (i)=k\), is fixed), where one specific element \(x_k\) can be mapped to any other of the \(n_B-1\) elements, and the remaining elements must not be mapped to their counterparts (\(\sigma (j) \ne j\)). Then there are \(D'(n_B-1)\) derangements \(\sigma \) realizing a specific mapping \(\sigma (i)=k\). Denoting the number of derangements of n elements by D(n), the sum above becomes
with \(D(n) = |{\mathfrak {S}}| = n! \sum _{k=0}^n (-1)^k / k!\) [67], and \(D'(n-1) = \sum _{k=0}^{n-1} (n-2)!(n-1-k)!(-1)^k\) by Proposition 4 below.
Given that \(|C_s| + |C_t|\) must cut at least \(n_B\) edges and that \(f_{bal}\) is increasing, \(f_{bal}\) is minimized if \(|C_s| = |C_t| = n_B/2\). As a result, if \(\beta \) is large enough such that \(f_{bal}\) dominates the cost, then a minimum cooperative cut in \(\mathcal {G}\) bisects the \(\mathcal {G}_B\) subgraph of \(\mathcal {G}\) optimally. \(\square \)
Proposition 4
Let \(D'(n)\) be the number of permutations of n elements where for one fixed element \(i'\) we allow \(\sigma (i') \in \{1, \ldots , n\}\), but \(\sigma (i) \ne i\) for all \(i \ne i'\). Then \(D'(n) = \sum _{k=0}^{n} \frac{(n-1)!}{k!}(n-k)! (-1)^k\).
Proof
\(D'(n)\) can be derived by the method of the forbidden board [67, pp. 71-73]. Let, without loss of generality, \(i'=n\), so the forbidden board is \(B = \{(1,1), (2,2), \ldots , (n-1,n-1)\}\). Let \(N_j\) be the number of permutations \(\sigma \) for which \(\big |\{(i,\sigma (i)\}_{i=1}^n \cap B\big | = j\), and let \(r_k\) be the number of k-subsets of B such that no two elements have a coordinate in common. Here, \(r_k~=~{n-1 \atopwithdelims ()k}\). Then \(D'(n) = N_0 = N_n(0)\) for
and hence \(D'(n) = \sum _{k=0}^{n} \frac{(n-1)!}{k!} (n-k)! (-1)^k\) (Fig. 8). \(\square \)
Convolutions of submodular functions are not always submodular
The non-submodularity of convolutions was mentioned already in Lovász [54]. For completeness, we show an explicit example that illustrates that non-submodularity also holds for the special case of polymatroidal flows.
Proposition 5
The convolution of two submodular functions \((f *g)(A) = \min _{B \subseteq A}f(B) + g(A{\setminus } B)\) is not in general submodular. In particular, this also holds for the cut cost functions occurring in the dual problems of polymatroidal maximum flows.
To show Proposition 5, consider the graph in Fig. 5 with a submodular edge cost function \(f(A) = \max _{e \in A}w(e)\). The two submodular functions that are convolved in the corresponding polymatroidal flow are the decompositions
Both \({\mathrm {cap}}^{\text {out}}\) and \({\mathrm {cap}}^{\text {in}}\) are submodular functions from \(2^\mathcal {E}\) to \({\mathbb {R}}_+\). Their convolution h is
For h to be submodular, it must satisfy the condition of diminishing marginal costs, i.e., for any e and \(A \subseteq B \subseteq \mathcal {E}{\setminus } e\), it must hold that \(h(e \mid A) \ge h(e \mid B)\). Now, let \(A = \{e_2\}\) and \(B = \{e_1, e_2\}\). The convolution here means to pair \(e_3\) either with \(e_1\) or \(e_2\). Then, if \(a < b\),
Hence, \(h(e_3 \mid A) < h(e_3 \mid B)\), disproving submodularity of h.
Cooperative cuts and polymatroidal networks
We next prove Lemma 6 that relates the approximation \(\hat{f}_{pf}\) to maxflow problems in polymatroidal networks.
Proof
(Lemma 6 ) The first step is the dual of the polymatroidal flow. Let \({\mathrm {cap}}^{\text {in}}: 2^\mathcal {E}\rightarrow {\mathbb {R}}_+\) be the joint incoming capacity, \({\mathrm {cap}}^{\text {in}}(C) = \sum _{v \in V}{\mathrm {cap}}_v^{\text {in}}(C \cap \delta ^-v)\), and let equivalently \({\mathrm {cap}}^{\text {out}}\) be the joint outgoing capacity. The dual of the polymatroidal maximum flow is a minimum cut problem whose cost is a convolution of edge capacities [54]:
We will relate this dual to the approximation \(\hat{f}_{pf}\). Given a minimal (s, t)-cut C, let \(\varPi (C)\) be a partition of C, and \(C_v^{\text {in}} = C^\varPi _v \cap \delta ^-_v\) and \(C_v^{\text {out}} = C^\varPi _v \cap \delta ^+_v\). The cut C partitions the nodes into two sets \(\mathcal {V}_s\) containing s and \(\mathcal {V}_t\) containing t. Since C is a minimal directed cut, it contains only edges from the s side \(\mathcal {V}_s\) to the t side \(\mathcal {V}_t\) of the graph. In consequence, \(C_v^{\text {in}} = \emptyset \) if v is on the s side, and \(C_v^{\text {out}} = \emptyset \) otherwise. Hence, \(C^{\text {in}}_v \cup C^{\text {out}}_v\) is equal to either \(C^{\text {in}}_v\) or \(C^{\text {out}}_v\), and since \(f(\emptyset )=0\), it holds that \(f(C^{\text {in}}_v \cup C^{\text {out}}_v) = f(C^{\text {in}}_v) + f(C^{\text {out}}_v)\). Then, starting with the definition of \(\hat{f}_{pf}\),
The minimum in Eq. (67) is taken over all feasible partitions \(\varPi (C)\) and their resulting intersections with the sets \(\delta ^+v, \delta ^-v\). Then we use the notation \(C^{\text {in}} = \bigcup _{v \in \mathcal {V}}C_v^{\text {in}}\) for all edges assigned to their head nodes, and \(C^{\text {out}} = \bigcup _{v \in \mathcal {V}}C_v^{\text {out}}\). The minima in Eqs. (69) and (70) are again taken over all partitions in \({\mathcal {P}}_C\). The final equality follows from the above definition of a convolution of submodular functions. \(\square \)
Rights and permissions
About this article
Cite this article
Jegelka, S., Bilmes, J.A. Graph cuts with interacting edge weights: examples, approximations, and algorithms. Math. Program. 162, 241–282 (2017). https://doi.org/10.1007/s10107-016-1038-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10107-016-1038-y