--- _id: '916' abstract: - lang: eng text: We study the quadratic assignment problem, in computer vision also known as graph matching. Two leading solvers for this problem optimize the Lagrange decomposition duals with sub-gradient and dual ascent (also known as message passing) updates. We explore this direction further and propose several additional Lagrangean relaxations of the graph matching problem along with corresponding algorithms, which are all based on a common dual ascent framework. Our extensive empirical evaluation gives several theoretical insights and suggests a new state-of-the-art anytime solver for the considered problem. Our improvement over state-of-the-art is particularly visible on a new dataset with large-scale sparse problem instances containing more than 500 graph nodes each. article_processing_charge: No author: - first_name: Paul full_name: Swoboda, Paul id: 446560C6-F248-11E8-B48F-1D18A9856A87 last_name: Swoboda - first_name: Carsten full_name: Rother, Carsten last_name: Rother - first_name: Carsten full_name: Abu Alhaija, Carsten last_name: Abu Alhaija - first_name: Dagmar full_name: Kainmueller, Dagmar last_name: Kainmueller - first_name: Bogdan full_name: Savchynskyy, Bogdan last_name: Savchynskyy citation: ama: 'Swoboda P, Rother C, Abu Alhaija C, Kainmueller D, Savchynskyy B. A study of lagrangean decompositions and dual ascent solvers for graph matching. In: Vol 2017. IEEE; 2017:7062-7071. doi:10.1109/CVPR.2017.747' apa: 'Swoboda, P., Rother, C., Abu Alhaija, C., Kainmueller, D., & Savchynskyy, B. (2017). A study of lagrangean decompositions and dual ascent solvers for graph matching (Vol. 2017, pp. 7062–7071). Presented at the CVPR: Computer Vision and Pattern Recognition, Honolulu, HA, United States: IEEE. https://doi.org/10.1109/CVPR.2017.747' chicago: Swoboda, Paul, Carsten Rother, Carsten Abu Alhaija, Dagmar Kainmueller, and Bogdan Savchynskyy. “A Study of Lagrangean Decompositions and Dual Ascent Solvers for Graph Matching,” 2017:7062–71. IEEE, 2017. https://doi.org/10.1109/CVPR.2017.747. ieee: 'P. Swoboda, C. Rother, C. Abu Alhaija, D. Kainmueller, and B. Savchynskyy, “A study of lagrangean decompositions and dual ascent solvers for graph matching,” presented at the CVPR: Computer Vision and Pattern Recognition, Honolulu, HA, United States, 2017, vol. 2017, pp. 7062–7071.' ista: 'Swoboda P, Rother C, Abu Alhaija C, Kainmueller D, Savchynskyy B. 2017. A study of lagrangean decompositions and dual ascent solvers for graph matching. CVPR: Computer Vision and Pattern Recognition vol. 2017, 7062–7071.' mla: Swoboda, Paul, et al. A Study of Lagrangean Decompositions and Dual Ascent Solvers for Graph Matching. Vol. 2017, IEEE, 2017, pp. 7062–71, doi:10.1109/CVPR.2017.747. short: P. Swoboda, C. Rother, C. Abu Alhaija, D. Kainmueller, B. Savchynskyy, in:, IEEE, 2017, pp. 7062–7071. conference: end_date: 2017-07-26 location: Honolulu, HA, United States name: 'CVPR: Computer Vision and Pattern Recognition' start_date: 2017-07-21 date_created: 2018-12-11T11:49:11Z date_published: 2017-01-01T00:00:00Z date_updated: 2023-09-26T15:41:40Z day: '01' ddc: - '000' department: - _id: VlKo doi: 10.1109/CVPR.2017.747 ec_funded: 1 external_id: isi: - '000418371407018' file: - access_level: open_access checksum: e38a2740daad1ea178465843b5072906 content_type: application/pdf creator: dernst date_created: 2019-01-18T12:49:38Z date_updated: 2020-07-14T12:48:15Z file_id: '5848' file_name: 2017_CVPR_Swoboda2.pdf file_size: 944332 relation: main_file file_date_updated: 2020-07-14T12:48:15Z has_accepted_license: '1' intvolume: ' 2017' isi: 1 language: - iso: eng month: '01' oa: 1 oa_version: Submitted Version page: 7062-7071 project: - _id: 25FBA906-B435-11E9-9278-68D0E5697425 call_identifier: FP7 grant_number: '616160' name: 'Discrete Optimization in Computer Vision: Theory and Practice' publication_identifier: isbn: - 978-153860457-1 publication_status: published publisher: IEEE publist_id: '6525' quality_controlled: '1' scopus_import: '1' status: public title: A study of lagrangean decompositions and dual ascent solvers for graph matching type: conference user_id: c635000d-4b10-11ee-a964-aac5a93f6ac1 volume: 2017 year: '2017' ... --- _id: '915' abstract: - lang: eng text: We propose a dual decomposition and linear program relaxation of the NP-hard minimum cost multicut problem. Unlike other polyhedral relaxations of the multicut polytope, it is amenable to efficient optimization by message passing. Like other polyhedral relaxations, it can be tightened efficiently by cutting planes. We define an algorithm that alternates between message passing and efficient separation of cycle- and odd-wheel inequalities. This algorithm is more efficient than state-of-the-art algorithms based on linear programming, including algorithms written in the framework of leading commercial software, as we show in experiments with large instances of the problem from applications in computer vision, biomedical image analysis and data mining. article_processing_charge: No author: - first_name: Paul full_name: Swoboda, Paul id: 446560C6-F248-11E8-B48F-1D18A9856A87 last_name: Swoboda - first_name: Bjoern full_name: Andres, Bjoern last_name: Andres citation: ama: 'Swoboda P, Andres B. A message passing algorithm for the minimum cost multicut problem. In: Vol 2017. IEEE; 2017:4990-4999. doi:10.1109/CVPR.2017.530' apa: 'Swoboda, P., & Andres, B. (2017). A message passing algorithm for the minimum cost multicut problem (Vol. 2017, pp. 4990–4999). Presented at the CVPR: Computer Vision and Pattern Recognition, Honolulu, HA, United States: IEEE. https://doi.org/10.1109/CVPR.2017.530' chicago: Swoboda, Paul, and Bjoern Andres. “A Message Passing Algorithm for the Minimum Cost Multicut Problem,” 2017:4990–99. IEEE, 2017. https://doi.org/10.1109/CVPR.2017.530. ieee: 'P. Swoboda and B. Andres, “A message passing algorithm for the minimum cost multicut problem,” presented at the CVPR: Computer Vision and Pattern Recognition, Honolulu, HA, United States, 2017, vol. 2017, pp. 4990–4999.' ista: 'Swoboda P, Andres B. 2017. A message passing algorithm for the minimum cost multicut problem. CVPR: Computer Vision and Pattern Recognition vol. 2017, 4990–4999.' mla: Swoboda, Paul, and Bjoern Andres. A Message Passing Algorithm for the Minimum Cost Multicut Problem. Vol. 2017, IEEE, 2017, pp. 4990–99, doi:10.1109/CVPR.2017.530. short: P. Swoboda, B. Andres, in:, IEEE, 2017, pp. 4990–4999. conference: end_date: 2017-07-26 location: Honolulu, HA, United States name: 'CVPR: Computer Vision and Pattern Recognition' start_date: 2017-07-21 date_created: 2018-12-11T11:49:11Z date_published: 2017-07-01T00:00:00Z date_updated: 2023-09-26T15:43:27Z day: '01' ddc: - '000' department: - _id: VlKo doi: 10.1109/CVPR.2017.530 ec_funded: 1 external_id: isi: - '000418371405009' file: - access_level: open_access checksum: 7e51dacefa693574581a32da3eff63dc content_type: application/pdf creator: dernst date_created: 2019-01-18T12:52:46Z date_updated: 2020-07-14T12:48:15Z file_id: '5849' file_name: Swoboda_A_Message_Passing_CVPR_2017_paper.pdf file_size: 883264 relation: main_file file_date_updated: 2020-07-14T12:48:15Z has_accepted_license: '1' intvolume: ' 2017' isi: 1 language: - iso: eng month: '07' oa: 1 oa_version: Submitted Version page: 4990-4999 project: - _id: 25FBA906-B435-11E9-9278-68D0E5697425 call_identifier: FP7 grant_number: '616160' name: 'Discrete Optimization in Computer Vision: Theory and Practice' publication_identifier: isbn: - 978-153860457-1 publication_status: published publisher: IEEE publist_id: '6526' quality_controlled: '1' scopus_import: '1' status: public title: A message passing algorithm for the minimum cost multicut problem type: conference user_id: c635000d-4b10-11ee-a964-aac5a93f6ac1 volume: 2017 year: '2017' ... --- _id: '917' abstract: - lang: eng text: We propose a general dual ascent framework for Lagrangean decomposition of combinatorial problems. Although methods of this type have shown their efficiency for a number of problems, so far there was no general algorithm applicable to multiple problem types. In this work, we propose such a general algorithm. It depends on several parameters, which can be used to optimize its performance in each particular setting. We demonstrate efficacy of our method on graph matching and multicut problems, where it outperforms state-of-the-art solvers including those based on subgradient optimization and off-the-shelf linear programming solvers. article_processing_charge: No author: - first_name: Paul full_name: Swoboda, Paul id: 446560C6-F248-11E8-B48F-1D18A9856A87 last_name: Swoboda - first_name: Jan full_name: Kuske, Jan last_name: Kuske - first_name: Bogdan full_name: Savchynskyy, Bogdan last_name: Savchynskyy citation: ama: 'Swoboda P, Kuske J, Savchynskyy B. A dual ascent framework for Lagrangean decomposition of combinatorial problems. In: Vol 2017. IEEE; 2017:4950-4960. doi:10.1109/CVPR.2017.526' apa: 'Swoboda, P., Kuske, J., & Savchynskyy, B. (2017). A dual ascent framework for Lagrangean decomposition of combinatorial problems (Vol. 2017, pp. 4950–4960). Presented at the CVPR: Computer Vision and Pattern Recognition, Honolulu, HA, United States: IEEE. https://doi.org/10.1109/CVPR.2017.526' chicago: Swoboda, Paul, Jan Kuske, and Bogdan Savchynskyy. “A Dual Ascent Framework for Lagrangean Decomposition of Combinatorial Problems,” 2017:4950–60. IEEE, 2017. https://doi.org/10.1109/CVPR.2017.526. ieee: 'P. Swoboda, J. Kuske, and B. Savchynskyy, “A dual ascent framework for Lagrangean decomposition of combinatorial problems,” presented at the CVPR: Computer Vision and Pattern Recognition, Honolulu, HA, United States, 2017, vol. 2017, pp. 4950–4960.' ista: 'Swoboda P, Kuske J, Savchynskyy B. 2017. A dual ascent framework for Lagrangean decomposition of combinatorial problems. CVPR: Computer Vision and Pattern Recognition vol. 2017, 4950–4960.' mla: Swoboda, Paul, et al. A Dual Ascent Framework for Lagrangean Decomposition of Combinatorial Problems. Vol. 2017, IEEE, 2017, pp. 4950–60, doi:10.1109/CVPR.2017.526. short: P. Swoboda, J. Kuske, B. Savchynskyy, in:, IEEE, 2017, pp. 4950–4960. conference: end_date: 2017-07-26 location: Honolulu, HA, United States name: 'CVPR: Computer Vision and Pattern Recognition' start_date: 2017-07-21 date_created: 2018-12-11T11:49:11Z date_published: 2017-07-01T00:00:00Z date_updated: 2023-09-26T15:41:11Z day: '01' ddc: - '000' department: - _id: VlKo doi: 10.1109/CVPR.2017.526 ec_funded: 1 external_id: isi: - '000418371405005' file: - access_level: open_access checksum: 72fd291046bd8e5717961bd68f6b6f03 content_type: application/pdf creator: dernst date_created: 2019-01-18T12:45:55Z date_updated: 2020-07-14T12:48:15Z file_id: '5847' file_name: 2017_CVPR_Swoboda.pdf file_size: 898652 relation: main_file file_date_updated: 2020-07-14T12:48:15Z has_accepted_license: '1' intvolume: ' 2017' isi: 1 language: - iso: eng month: '07' oa: 1 oa_version: Submitted Version page: 4950-4960 project: - _id: 25FBA906-B435-11E9-9278-68D0E5697425 call_identifier: FP7 grant_number: '616160' name: 'Discrete Optimization in Computer Vision: Theory and Practice' publication_identifier: isbn: - 978-153860457-1 publication_status: published publisher: IEEE publist_id: '6524' quality_controlled: '1' scopus_import: '1' status: public title: A dual ascent framework for Lagrangean decomposition of combinatorial problems type: conference user_id: c635000d-4b10-11ee-a964-aac5a93f6ac1 volume: 2017 year: '2017' ... --- _id: '274' abstract: - lang: eng text: We consider the problem of estimating the partition function Z(β)=∑xexp(−β(H(x)) of a Gibbs distribution with a Hamilton H(⋅), or more precisely the logarithm of the ratio q=lnZ(0)/Z(β). It has been recently shown how to approximate q with high probability assuming the existence of an oracle that produces samples from the Gibbs distribution for a given parameter value in [0,β]. The current best known approach due to Huber [9] uses O(qlnn⋅[lnq+lnlnn+ε−2]) oracle calls on average where ε is the desired accuracy of approximation and H(⋅) is assumed to lie in {0}∪[1,n]. We improve the complexity to O(qlnn⋅ε−2) oracle calls. We also show that the same complexity can be achieved if exact oracles are replaced with approximate sampling oracles that are within O(ε2qlnn) variation distance from exact oracles. Finally, we prove a lower bound of Ω(q⋅ε−2) oracle calls under a natural model of computation. article_processing_charge: No author: - first_name: Vladimir full_name: Kolmogorov, Vladimir id: 3D50B0BA-F248-11E8-B48F-1D18A9856A87 last_name: Kolmogorov citation: ama: 'Kolmogorov V. A faster approximation algorithm for the Gibbs partition function. In: Proceedings of the 31st Conference On Learning Theory. Vol 75. ML Research Press; 2017:228-249.' apa: Kolmogorov, V. (2017). A faster approximation algorithm for the Gibbs partition function. In Proceedings of the 31st Conference On Learning Theory (Vol. 75, pp. 228–249). ML Research Press. chicago: Kolmogorov, Vladimir. “A Faster Approximation Algorithm for the Gibbs Partition Function.” In Proceedings of the 31st Conference On Learning Theory, 75:228–49. ML Research Press, 2017. ieee: V. Kolmogorov, “A faster approximation algorithm for the Gibbs partition function,” in Proceedings of the 31st Conference On Learning Theory, 2017, vol. 75, pp. 228–249. ista: 'Kolmogorov V. 2017. A faster approximation algorithm for the Gibbs partition function. Proceedings of the 31st Conference On Learning Theory. COLT: Annual Conference on Learning Theory vol. 75, 228–249.' mla: Kolmogorov, Vladimir. “A Faster Approximation Algorithm for the Gibbs Partition Function.” Proceedings of the 31st Conference On Learning Theory, vol. 75, ML Research Press, 2017, pp. 228–49. short: V. Kolmogorov, in:, Proceedings of the 31st Conference On Learning Theory, ML Research Press, 2017, pp. 228–249. conference: end_date: 2018-07-09 name: 'COLT: Annual Conference on Learning Theory ' start_date: 2018-07-06 date_created: 2018-12-11T11:45:33Z date_published: 2017-12-27T00:00:00Z date_updated: 2023-10-17T12:32:13Z day: '27' ddc: - '510' department: - _id: VlKo ec_funded: 1 external_id: arxiv: - '1608.04223' file: - access_level: open_access checksum: 89db06a0e8083524449cb59b56bf4e5b content_type: application/pdf creator: dernst date_created: 2020-05-12T09:23:27Z date_updated: 2020-07-14T12:45:45Z file_id: '7820' file_name: 2018_PMLR_Kolmogorov.pdf file_size: 408974 relation: main_file file_date_updated: 2020-07-14T12:45:45Z has_accepted_license: '1' intvolume: ' 75' language: - iso: eng license: https://creativecommons.org/licenses/by/4.0/ month: '12' oa: 1 oa_version: Published Version page: 228-249 project: - _id: 25FBA906-B435-11E9-9278-68D0E5697425 call_identifier: FP7 grant_number: '616160' name: 'Discrete Optimization in Computer Vision: Theory and Practice' publication: Proceedings of the 31st Conference On Learning Theory publication_status: published publisher: ML Research Press publist_id: '7628' quality_controlled: '1' status: public title: A faster approximation algorithm for the Gibbs partition function tmp: image: /images/cc_by.png legal_code_url: https://creativecommons.org/licenses/by/4.0/legalcode name: Creative Commons Attribution 4.0 International Public License (CC-BY 4.0) short: CC BY (4.0) type: conference user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 volume: 75 year: '2017' ... --- _id: '5561' abstract: - lang: eng text: 'Graph matching problems as described in "Active Graph Matching for Automatic Joint Segmentation and Annotation of C. Elegans." by Kainmueller, Dagmar and Jug, Florian and Rother, Carsten and Myers, Gene, MICCAI 2014. Problems are in OpenGM2 hdf5 format (see http://hciweb2.iwr.uni-heidelberg.de/opengm/) and a custom text format used by the feature matching solver described in "Feature Correspondence via Graph Matching: Models and Global Optimization." by Lorenzo Torresani, Vladimir Kolmogorov and Carsten Rother, ECCV 2008, code at http://pub.ist.ac.at/~vnk/software/GraphMatching-v1.02.src.zip. ' acknowledgement: We thank Vladimir Kolmogorov and Stephan Saalfeld forinspiring discussions. article_processing_charge: No author: - first_name: Dagmar full_name: Kainmueller, Dagmar last_name: Kainmueller - first_name: Florian full_name: Jug, Florian last_name: Jug - first_name: Carsten full_name: Rother, Carsten last_name: Rother - first_name: Gene full_name: Meyers, Gene last_name: Meyers citation: ama: Kainmueller D, Jug F, Rother C, Meyers G. Graph matching problems for annotating C. Elegans. 2017. doi:10.15479/AT:ISTA:57 apa: Kainmueller, D., Jug, F., Rother, C., & Meyers, G. (2017). Graph matching problems for annotating C. Elegans. Institute of Science and Technology Austria. https://doi.org/10.15479/AT:ISTA:57 chicago: Kainmueller, Dagmar, Florian Jug, Carsten Rother, and Gene Meyers. “Graph Matching Problems for Annotating C. Elegans.” Institute of Science and Technology Austria, 2017. https://doi.org/10.15479/AT:ISTA:57. ieee: D. Kainmueller, F. Jug, C. Rother, and G. Meyers, “Graph matching problems for annotating C. Elegans.” Institute of Science and Technology Austria, 2017. ista: Kainmueller D, Jug F, Rother C, Meyers G. 2017. Graph matching problems for annotating C. Elegans, Institute of Science and Technology Austria, 10.15479/AT:ISTA:57. mla: Kainmueller, Dagmar, et al. Graph Matching Problems for Annotating C. Elegans. Institute of Science and Technology Austria, 2017, doi:10.15479/AT:ISTA:57. short: D. Kainmueller, F. Jug, C. Rother, G. Meyers, (2017). datarep_id: '57' date_created: 2018-12-12T12:31:32Z date_published: 2017-02-13T00:00:00Z date_updated: 2024-02-21T13:46:31Z day: '13' ddc: - '000' department: - _id: VlKo doi: 10.15479/AT:ISTA:57 file: - access_level: open_access checksum: 3dc3e1306a66028a34181ebef2923139 content_type: application/zip creator: system date_created: 2018-12-12T13:02:54Z date_updated: 2020-07-14T12:47:03Z file_id: '5614' file_name: IST-2017-57-v1+1_wormMatchingProblems.zip file_size: 327042819 relation: main_file file_date_updated: 2020-07-14T12:47:03Z has_accepted_license: '1' keyword: - graph matching - feature matching - QAP - MAP-inference license: https://creativecommons.org/publicdomain/zero/1.0/ month: '02' oa: 1 oa_version: Published Version publisher: Institute of Science and Technology Austria status: public title: Graph matching problems for annotating C. Elegans tmp: image: /images/cc_0.png legal_code_url: https://creativecommons.org/publicdomain/zero/1.0/legalcode name: Creative Commons Public Domain Dedication (CC0 1.0) short: CC0 (1.0) type: research_data user_id: 2DF688A6-F248-11E8-B48F-1D18A9856A87 year: '2017' ...