Corpus ID: 11811078

A Multi-Plane Block-Coordinate Frank-Wolfe Algorithm for Structural SVMs with a Costly max-Oracle

@article{Shah2014AMB,
  title={A Multi-Plane Block-Coordinate Frank-Wolfe Algorithm for Structural SVMs with a Costly max-Oracle},
  author={N. Shah and V. Kolmogorov and Christoph H. Lampert},
  journal={ArXiv},
  year={2014},
  volume={abs/1408.6804}
}
Structural support vector machines (SSVMs) are amongst the best performing models for many structured computer vision tasks, such as semantic image segmentation or human pose estimation. Training SSVMs, however, is computationally costly, since it requires repeated calls to a structured prediction subroutine (called max-oracle), which requires solving an optimization problem itself, e.g. a graph cut. In this work, we introduce a new technique for SSVM training that is more efficient than… Expand

References

SHOWING 1-10 OF 27 REFERENCES
Max-Margin Markov Networks
  • 1,461
  • PDF
An experimental comparison of min-cut/max- flow algorithms for energy minimization in vision
  • 3,393
  • PDF
Cutting-plane training of structural SVMs
  • 1,072
  • Highly Influential
  • PDF
Learning CRFs Using Graph Cuts
  • 311
  • PDF
On Parameter Learning in CRF-Based Approaches to Object Class Image Segmentation
  • 94
  • PDF
Large Margin Methods for Structured and Interdependent Output Variables
  • 2,198
  • PDF
Kernelized structural SVM learning for supervised object segmentation
  • 108
  • PDF
Structural SVM for visual localization and continuous state estimation
  • 37
  • PDF
Bundle Methods for Regularized Risk Minimization
  • 261
  • PDF
Making large-scale support vector machine learning practical
  • 1,819
  • Highly Influential
...
1
2
3
...