A Warning Propagation-Based Linear-Time-and-Space Algorithm for the Minimum Vertex Cover Problem on Giant Graphs

The presentation slides of the paper "Hong Xu, Kexuan Sun, Sven Koenig, and T. K. Satish Kumar. A warning propagation-based linear-time-and-space algorithm for the minimum vertex cover problem on giant graphs. In Proceedings of the 15th International Conference on the Integration of Constraint Programming, Artificial Intelligence, and Operations Research (CPAIOR), 567–584. 2018."

More details: http://www.hong.me/papers/xu2018b.html
Link to published paper:
https://doi.org/10.1007/978-3-319-93031-2_41

Transcript

1. A Warning Propagation-Based Linear-Time-and-Space Algorithm for the Minimum Vertex Cover Problem on Giant Graphs Hong Xu Kexuan Sun Sven Koenig T. K. Satish Kumar {hongx, kexuansu, skoenig}@usc.edu tkskwork@gmail.com June 27, 2018 University of Southern California, Los Angeles, California 90089, the United States of America The 15th International Conference on the Integration of Constraint Programming, Artificial Intelligence, and Operations Research (CPAIOR) Delft, The Netherlands
2. Summary The minimum vertex cover (MVC) problem is a classical computer science problem. Local search algorithms often require a good starting state. For giant graphs, it is desirable to develop a linear-time-and-space algorithm to find a vertex cover as small as possible. We developed MVC-WP, a family of warning propagation-based linear-time-and-space algorithms. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 1 / 25
3. Agenda Motivation MVC-WP: Warning Propagation-Based Linear-Time-and-Space Algorithms Experimental Evaluation Conclusion and Future Work Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 2 / 25
4. Agenda Motivation MVC-WP: Warning Propagation-Based Linear-Time-and-Space Algorithms Experimental Evaluation Conclusion and Future Work
5. Motivation: The Minimum Vertex Cover (MVC) Problem The minimum vertex cover (MVC) problem is a classical problem in computer science. On a graph G = V, E , a vertex cover (VC) is a subset of vertices S ⊆ V such that every edge in G has at least one endpoint vertex in S. The MVC problem is to find a VC of minimum cardinality. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 3 / 25
6. Motivation: Applications Applications: Computer network security (Filiol et al. 2007) Crew scheduling (Sherali et al. 1984) Construction of phylogenetic tree (Abu-Khzam et al. 2004) Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 4 / 25
7. Motivation: A Linear Algorithm for the MVC Problem For giant graphs: Exact algorithms in general do not scale well. Local search algorithms require a small VC as a starting state (Andrade et al. 2012; Cai 2015; Cai et al. 2013; Pullan 2009). A linear-time-and-space algorithm to construct a small VC is desirable. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 5 / 25
8. Agenda Motivation MVC-WP: Warning Propagation-Based Linear-Time-and-Space Algorithms Experimental Evaluation Conclusion and Future Work
9. The Warning Propagation (WP) Algorithm for the MVC Problem An iterative algorithm proposed and has only been used for theoretical analysis of properties of MVC on infinite graphs by (Weigt et al. 2006). u v 0 1 0 0 (a) u sends a message of 0 to v since one of its incoming messages from other vertices is 1. u v 1 0 0 0 (b) u sends a message of 1 to v since all other incoming messages are 0. There is a message of either 0 or 1 along each direction of each edge. Assume that u and v are two adjacent vertices. u sends v a message of 1 to “warn” v to indicate that u will not be selected in the vertex cover. Otherwise, u sends v a message of 0. Update all messages until convergence. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 6 / 25
10. Basic Idea Observation: The warning propagation (WP) algorithm does not work well in practice (slow to converge, output often not good). But, each iteration uses only linear amount of time and space. Basic idea of a linear-time-and-space algorithm: Run a few iterations of warning propagation and extract a vertex cover afterwards. However, can we improve the output by having some processing before and after warning propagation iterations? Yes. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 7 / 25
11. Major Steps of MVC-WP MVC-WP: Warning Propagation-Based Linear-Time-and-Space Algorithms Prune Leaves Initialize Messages Run Warning Propagation Iterations Remove Redundant Vertices Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 8 / 25
12. Major Steps of MVC-WP MVC-WP: Warning Propagation-Based Linear-Time-and-Space Algorithms Prune Leaves Initialize Messages Run Warning Propagation Iterations Remove Redundant Vertices
13. Prune Leaves A leaf vertex is a vertex of degree 1. Idea: Adding a leaf vertex into the vertex cover is no better than adding its neighbor. Adding the blue vertex to the vertex cover is at least as good as adding the red vertex. Therefore, we can remove the two colored vertices from the graph. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 9 / 25
14. Major Steps of MVC-WP MVC-WP: Warning Propagation-Based Linear-Time-and-Space Algorithms Prune Leaves Initialize Messages Run Warning Propagation Iterations Remove Redundant Vertices
15. Initialize Messages: Assuming Random Graph Models The standard warning propagation algorithm initialize all messages to zero, but we should be able to do better. We treat the pruned graph as if it were generated by a random graph model (but the input graph itself is not required to be generated by a random graph model). Erdős-Rényi: Vertex degrees follow a Poisson distribution (P(d) ∝ cd/d!). Scale-free: Vertex degrees follow a power law (P(d) ∝ d−λ). Basic idea: Initialize messages based on the values of the messages when the warning propagation algorithm converges on an infinitely large random graph from analytical results. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 10 / 25
16. The Warning Propagation (WP) Algorithm for the MVC Problem An iterative algorithm proposed and has only been used for theoretical analysis of properties of MVC on infinite graphs by (Weigt et al. 2006). u v 0 1 0 0 (a) u sends a message of 0 to v since one of its incoming messages from other vertices is 1. u v 1 0 0 0 (b) u sends a message of 1 to v since all other incoming messages are 0. There is a message of either 0 or 1 along each direction of each edge. Assume that u and v are two adjacent vertices. u sends v a message of 1 to “warn” v to indicate that u will not be selected in the vertex cover. Otherwise, u sends v a message of 0. Update all messages until convergence. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 11 / 25
17. Initialize Messages: Convergent Message Values on Infinite Graphs On a given infinitely large random graph, let p0 denote the probability that a message is zero and assume that p0 is constant for all messages. u v 1 − p0 = ∞ d=1 pd−1 0 P(d) Erdős-Rényi: p0 = 1 − W(c)/c (Weigt et al. 2006) scale-free: p0 = (ζ(λ) − 1)/(ζ(λ) + 1 2λ ) Randomly initialize messages according to the calculated p0 . Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 12 / 25
18. Major Steps of MVC-WP MVC-WP: Warning Propagation-Based Linear-Time-and-Space Algorithms Prune Leaves Initialize Messages Run Warning Propagation Iterations Remove Redundant Vertices
19. Run Warning Propagation Iterations 0 1 2 3 4 5 M 228000 230000 232000 234000 236000 238000 240000 242000 244000 Average VC Size MVC-WP-ER MVC-WP-SF M: Number of iterations How many iterations? We choose M = 3: A small number but has fairly good effectiveness. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 13 / 25
20. Major Steps of MVC-WP MVC-WP: Warning Propagation-Based Linear-Time-and-Space Algorithms Prune Leaves Initialize Messages Run Warning Propagation Iterations Remove Redundant Vertices
21. Remove Redundant Vertices Proposed by (Cai 2015). Starting from a vertex cover, for each edge, if it has both endpoint vertices in the vertex cover, mark them as removable. Recursively remove removable vertices and update marks of other vertices accordingly. Guaranteed to output a minimal vertex cover. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 14 / 25
22. Agenda Motivation MVC-WP: Warning Propagation-Based Linear-Time-and-Space Algorithms Experimental Evaluation Conclusion and Future Work
23. Benchmark Instances Network repository (http://networkrepository.com/) misc networks (397) web networks (18) brain networks (26) Street networks (8) (http: //www.cc.gatech.edu/dimacs10/archive/streets.shtml) All graphs have more than 100,000 vertices. We have also converted these graphs in various formats to the DIMACS format: http://files.hong.me/papers/xu2018b-data Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 15 / 25
24. Algorithms Our Algorithms: MVC-WP-ER: Assuming Erdős-Rényi random graphs. MVC-WP-SF: Assuming scale-free random graphs. Competitors: MVC-2 (Vazirani 2003) ConstructVC (Cai 2015) R (Andrade et al. 2012) MVC-MPL (Xu et al. 2017) MVC-L (Xu et al. 2017) To ensure fair comparison, leaf pruning and redundant vertex removal are also applied to each algorithm. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 16 / 25
25. Results Our Alternative Misc Web Street Brain Algorithm Algorithm (397) (18) (8) (26) MVC-WP-ER ConstructVC 211/39/147 12/1/5 8/0/0 0/0/26 MVC-2 241/46/110 16/1/1 8/0/0 26/0/0 R 376/16/5 17/1/0 8/0/0 26/0/0 MVC-MPL 317/18/62 17/1/0 1/0/7 26/0/0 MVC-L 364/19/14 17/1/0 8/0/0 26/0/0 MVC-WP-SF ConstructVC 209/38/150 11/1/6 8/0/0 0/0/26 MVC-2 249/45/103 15/1/2 8/0/0 26/0/0 R 377/15/5 17/1/0 8/0/0 26/0/0 MVC-MPL 316/18/63 17/1/0 1/0/7 26/0/0 MVC-L 363/21/13 17/1/0 8/0/0 26/0/0 Comparison of sizes of vertex covers produced by MVC-WP-ER and MVC-WP-SF, respectively, with those of alternative algorithms. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 17 / 25
26. Results: Misc Networks 10 9 8 7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10 (VC Size of MVC-WP-ER - VC Size of an Alternative Algorithm) / Larger VC Size (%) 0 50 100 150 Number of Benchmark Instances ConstructVC MVC-2 R MVC-L MVC-MPL (a) MVC-WP-ER versus ConstructVC/MVC-2/R/MVC-L/MVC-MPL 10 9 8 7 6 5 4 3 2 1 0 1 2 3 4 5 6 7 8 9 10 (VC Size of MVC-WP-SF - VC Size of an Alternative Algorithm) / Larger VC Size (%) 0 50 100 150 Number of Benchmark Instances ConstructVC MVC-2 R MVC-L MVC-MPL (b) MVC-WP-SF versus ConstructVC/MVC-2/R/MVC-L/MVC-MPL Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 18 / 25
27. Results: Web Networks Instance ConstructVC MVC-2 R MVC-L MVC-MPL MVC-WP-ER MVC-WP-SF web-wikipedia-link-it 991,272 987,621 1,039,011 1,018,672 1,020,827 972,275 972,670 web-wikipedia-growth 914,746 926,530 966,410 950,741 940,302 909,910 909,989 web-BerkStan 5,542 5,469 5,605 5,567 5,726 5,463 5,469 web-italycnr-2000 99,645 99,609 110,559 104,272 103,153 97,844 97,932 web-uk-2005 127,774 127,774 127,774 127,774 127,774 127,774 127,774 web-Stanford 126,603 126,960 136,048 130,248 128,296 123,540 123,890 web-BerkStan-dir 290,206 291,277 320,384 310,483 304,623 285,593 285,934 web-google-dir 350,676 355,930 387,462 379,508 365,427 351,241 351,343 web-wikipedia2009 650,888 654,152 657,343 656,131 655,813 652,241 652,363 web-it-2004 415,408 415,083 415,915 415,533 415,042 414,835 414,972 web-wikipedia-link-fr 1,574,973 1,558,998 1,626,052 1,598,021 1,597,887 1,538,658 1,538,960 web-hudong 503,373 504,025 506,598 505,903 504,839 503,335 503,359 web-arabic-2005 114,504 114,743 115,161 114,999 115,004 114,721 114,727 web-baidu-baike 637,805 638,538 640,537 639,811 639,935 637,796 637,815 web-NotreDame 76,468 76,257 80,341 79,013 77,893 75,735 75,953 web-sk-2005 58,238 58,300 58,669 58,443 58,370 58,347 58,349 web-wiki-ch-internal 260,354 260,476 261,571 261,244 260,927 260,213 260,231 web-baidu-baike-related 144,388 146,588 151,689 149,957 148,749 145,272 145,249 Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 19 / 25
28. Results: Street Networks Instance ConstructVC MVC-2 R MVC-L MVC-MPL MVC-WP-ER MVC-WP-SF asia 6,087,218 6,099,227 6,130,265 6,104,699 6,018,875 6,053,077 6,049,489 germany 5,822,566 5,834,966 5,864,005 5,841,507 5,768,621 5,792,165 5,789,947 great-britain 3,837,647 3,843,980 3,857,098 3,844,972 3,804,317 3,821,741 3,820,618 luxembourg 58,168 58,267 58,456 58,230 57,417 57,823 57,810 belgium 739,185 741,647 747,374 742,968 729,190 733,264 732,877 netherlands 1,133,606 1,141,977 1,147,202 1,141,786 1,131,859 1,127,358 1,125,978 italy 3,425,723 3,434,538 3,452,907 3,434,830 3,374,512 3,401,824 3,400,005 europe 25,903,178 25,968,573 26,104,371 25,983,279 25,589,132 25,743,670 25,730,398 Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 20 / 25
29. Results: Brain Networks Instance ConstructVC MVC-2 R MVC-L MVC-MPL MVC-WP-ER MVC-WP-SF 0025871-session-1-bg 688,391 695,636 701,228 698,818 699,234 694,616 694,625 0025872-session-2-bg 706,691 714,407 720,128 717,692 718,017 713,557 713,599 0025869-session-1-bg 629,715 637,159 642,647 640,327 641,147 636,057 636,050 0025876-session-1-bg 712,322 721,476 728,763 725,679 726,560 720,332 720,404 0025865-session-2-bg 656,483 663,359 669,205 666,846 666,996 662,446 662,564 0025868-session-1-bg 662,749 670,415 676,703 674,029 674,729 669,473 669,452 0025872-session-1-bg 692,476 700,076 705,979 703,495 704,039 699,263 699,284 0025864-session-2-bg 631,361 638,668 644,392 641,875 642,737 637,833 637,831 0025912-session-2 716,154 724,000 730,201 727,577 727,982 723,070 723,165 0025868-session-2-bg 662,025 669,661 676,133 673,415 673,972 668,845 668,784 0025869-session-2-bg 651,656 659,116 665,348 662,796 662,931 658,201 658,171 0025873-session-1-bg 595,166 601,378 605,916 604,042 604,270 600,488 600,467 0025870-session-2-bg 735,348 744,205 751,736 748,478 749,706 743,052 743,103 0025865-session-1-bg 676,416 683,459 689,108 686,788 687,193 682,528 682,526 0025889-session-2 674,887 683,034 689,378 686,679 687,462 682,117 682,056 0025876-session-2-bg 701,544 710,627 717,965 714,784 715,793 709,579 709,604 0025867-session-1-bg 682,118 689,846 695,806 693,294 693,749 688,861 688,829 0025874-session-2-bg 701,559 709,752 716,330 713,553 714,320 708,796 708,786 0025873-session-2-bg 635,363 641,971 647,566 645,325 645,449 641,349 641,441 0025889-session-1 644,858 651,812 657,483 655,226 655,525 651,102 651,113 0025870-session-1-bg 721,906 730,773 738,123 734,922 736,193 729,761 729,805 0025871-session-2-bg 674,478 681,551 687,309 684,955 685,328 680,690 680,686 0025864-session-1-bg 634,702 642,146 647,884 645,431 646,130 641,119 641,058 0025867-session-2-bg 673,075 680,577 686,239 683,853 684,641 679,654 679,647 0025878-session-1-bg 636,617 644,044 650,351 647,661 648,109 643,237 643,190 0025886-session-1 713,597 721,617 728,101 725,332 726,009 720,641 720,638 Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 21 / 25
30. Results: Small Graphs (DIMACS Competition) Graph Algorithm Instance |V| MVC ConstructVC MVC-2 R MVC-L MVC-MPL MVC-ER MVC-SF c-fat200-1 200 188 188 188 189 189 189 188 190 c-fat200-2 200 176 176 176 177 177 177 177 177 c-fat200-5 200 142 142 142 143 142 143 142 142 c-fat500-10 500 374 374 374 374 375 375 374 374 c-fat500-1 500 486 486 486 487 487 487 487 488 c-fat500-2 500 474 474 474 474 475 474 474 474 c-fat500-5 500 436 436 436 437 437 437 437 436 hamming10-2 1024 512 760 674 787 741 738 593 651 hamming6-2 64 32 45 40 44 40 39 34 41 hamming6-4 64 60 60 60 60 60 60 60 60 hamming8-2 256 128 155 185 188 182 178 141 160 hamming8-4 256 240 250 248 248 246 246 245 248 johnson16-2-4 120 112 112 112 112 112 112 112 112 johnson8-2-4 28 24 24 24 24 24 24 24 24 johnson8-4-4 70 56 56 60 61 60 60 58 56 keller4 171 160 162 164 163 163 163 163 161 p-hat300-1 300 292 295 294 295 294 294 295 295 p-hat300-2 300 275 279 285 286 286 284 280 282 p-hat300-3 300 264 270 273 278 278 277 273 271 p-hat500-1 500 491 494 494 494 494 494 493 495 p-hat500-2 500 464 470 479 482 480 481 472 473 san200-0 200 170 184 185 185 185 185 185 185 san200-0 200 130 155 155 159 156 154 154 155 san200-0 200 140 163 163 168 167 166 164 165 san200-0 200 156 169 169 173 172 171 171 172 sanr200-0 200 182 187 188 187 188 187 187 188 Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 22 / 25
31. Running Time MVC-WP-ER and MVC-WP-SF are roughly a few times slower than the others. But 80% of their runs terminated within 300ms. A detailed comparison is difficult, e.g., reading input files took a few hundreds times longer than actually solving them. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 23 / 25
32. Is the Message Initialization Useful? MVC-WP-1: The standard message initialization, i.e., all messages are 0. < -8 -8 -6 -4 -2 0 2 4 6 8 > 8 (VC Size of MVC-WP - VC Size of MVC-WP-1) / Larger VC Size (%) 0 50 100 150 200 250 Number of Benchmark Instances MVC-WP-ER MVC-WP-SF Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 24 / 25
33. Agenda Motivation MVC-WP: Warning Propagation-Based Linear-Time-and-Space Algorithms Experimental Evaluation Conclusion and Future Work
34. Conclusion and Future Work We need a linear-time-and-space algorithm for the MVC problem, because local search algorithms require a good starting vertex cover. We developed MVC-WP, warning propagation-based linear-time-and-space algorithms that find small vertex covers for giant graphs. Our experimental results showed that MVC-WP outperformed other competitors. We have compiled various sets of graphs in many different formats into a set of giant graphs in the DIMACS format: http://files.hong.me/papers/xu2018b-data (Future work) Combine MVC-WP with local search algorithms. Apply similar techniques on other combinatorial problems. Xu et al. (University of Southern California) A Warning Propagation-Based Linear-Time-and-Space Algorithm for the MVC Problem 25 / 25
35. References I F. N. Abu-Khzam, R. L. Collins, M. R. Fellows, M. A. Langston, W. H. Suters, and C. T. Symons. “Kernelization Algorithms for the Vertex Cover Problem: Theory and Experiments”. In: the Workshop on Algorithm Engineering and Experiments. 2004. D. V. Andrade, M. G. C. Resende, and R. F. Werneck. “Fast local search for the maximum independent set problem”. In: Journal of Heuristics 18.4 (2012), pp. 525–547. doi: 10.1007/s10732-012-9196-4. S. Cai. “Balance between Complexity and Quality: Local Search for Minimum Vertex Cover in Massive Graphs”. In: the International Joint Conference on Artificial Intelligence. 2015, pp. 747–753. S. Cai, K. Su, C. Luo, and A. Sattar. “NuMVC: An Efficient Local Search Algorithm for Minimum Vertex Cover”. In: Journal of Artificial Intelligence Research 46.1 (2013), pp. 687–716. É. Filiol, E. Franc, A. Gubbioli, B. Moquet, and G. Roblot. “Combinatorial Optimisation of Worm Propagation on an Unknown Network”. In: International Journal of Computer, Electrical, Automation, Control and Information Engineering 1.10 (2007), pp. 2931–2937.
36. References II W. Pullan. “Optimisation of unweighted/weighted maximum independent sets and minimum vertex covers”. In: Discrete Optimization 6.2 (2009), pp. 214–219. doi: 10.1016/j.disopt.2008.12.001. H. D. Sherali and M. Rios. “An Air Force Crew Allocation and Scheduling Problem”. In: The Journal of the Operational Research Society 35.2 (1984), pp. 91–103. V. V. Vazirani. Approximation Algorithms. Springer, 2003. M. Weigt and H. Zhou. “Message passing for vertex covers”. In: Physical Review E 74.4 (2006), p. 046110. doi: 10.1103/PhysRevE.74.046110. H. Xu, T. K. S. Kumar, and S. Koenig. “A Linear-Time and Linear-Space Algorithm for the Minimum Vertex Cover Problem on Giant Graphs”. In: the International Symposium on Combinatorial Search. 2017, pp. 173–174.