Research map / Optimization / Nonconvex and smooth optimization
Local Submodularization for Binary Pairwise Energies
Lena Gorelick, Yuri Y. Boykov, Olga Veksler, Ismail Ben Ayed, Andrew Delong
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use LSA-TR instead of truncating non-submodular terms | truncation · Chinese characters inpainting | Mean energy: -16089.2 -> -49547.61 | Improved |
| use LSA-TR instead of TRWS | TRWS · Chinese characters inpainting | Mean energy: -49496.84 -> -49547.61 | Improved |
| use LSA-TR instead of MCBC for speed | MCBC · Chinese characters inpainting | Mean runtime: 2053.89 sec -> 0.21 sec | Improved |
| use LSA-TR instead of MCBC for energy | MCBC · Chinese characters inpainting | Mean energy: -49550.1 -> -49547.61 | Hurt |
| use LSA-AUX instead of LSA-TR (ablation) | LSA-TR · Chinese characters inpainting | Mean energy: -49547.61 -> -49515.95 | Hurt |
| add LSA-AUX-P variant over LSA-AUX (ablation) | LSA-AUX · Chinese characters inpainting | Mean runtime: 0.30 sec -> 0.16 sec | Improved |
| use purple-plane linear upper bounds for (0,0) and (1,1) instead of random orange/green planes | permutation bounds · binary pairwise energies | Optimization quality | Improved |
| use LSA-TR instead of linearized LSA-TR-L on submodular pairwise terms | LSA-TR-L · binary pairwise energies | Energy | Improved |
| use LSA-TR/LSA-AUX instead of QPBO, TRWS, LBP when curvature weight rises from 19 to 21 | QPBO, TRWS, LBP · angiogram curvature | Segmentation quality | Improved |
| raise curvature weight from 0.1 to 0.5 or 2 (ablation) | all optimizers · circle curvature toy | Energy | Mixed |
Closest papers by meaning
- Adaptive and Move Making Auxiliary Cuts for Binary Pairwise Energies (CVPR 2017 · similarity 0.91)
- DOPE: Distributed Optimization for Pairwise Energies (CVPR 2017 · similarity 0.88)
- Complexity of Discrete Energy Minimization Problems (ECCV 2016 · similarity 0.88)
- Sublabel-Accurate Relaxation of Nonconvex Energies (CVPR 2016 · similarity 0.87)
- Optimal approximation for unconstrained non-submodular minimization (ICML 2020 · similarity 0.86)
- Sublabel-Accurate Convex Relaxation of Vectorial Multilabel Energies (ECCV 2016 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
