A study of lagrangean decompositions and dual ascent solvers for graph matching

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.

Download
OA 2017_CVPR_Swoboda2.pdf 944.33 KB [Submitted Version]

Conference Paper | Published | English

Scopus indexed
Author
Swoboda, PaulISTA; Rother, Carsten; Abu Alhaija, Carsten; Kainmueller, Dagmar; Savchynskyy, Bogdan

Corresponding author has ISTA affiliation

Department
Abstract
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.
Publishing Year
Date Published
2017-01-01
Publisher
IEEE
Volume
2017
Page
7062-7071
Conference
CVPR: Computer Vision and Pattern Recognition
Conference Location
Honolulu, HA, United States
Conference Date
2017-07-21 – 2017-07-26
IST-REx-ID
916

Cite this

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
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
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.
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.
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.
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.
All files available under the following license(s):
Copyright Statement:
This Item is protected by copyright and/or related rights. [...]
Main File(s)
File Name
Access Level
OA Open Access
Date Uploaded
2019-01-18
MD5 Checksum
e38a2740daad1ea178465843b5072906


Export

Marked Publications

Open Data ISTA Research Explorer

Web of Science

View record in Web of Science®

Search this title in

Google Scholar
ISBN Search