Research map / Learning theory and algorithms / Combinatorial and search optimization
Maximum Persistency via Iterative Relaxed Inference in Graphical Models
Alexander Shekhovtsov, Paul Swoboda, Bogdan Savchynskyy
Read the paperCheck your idea against 137,217 papers
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 |
|---|---|---|---|
| add speedups (reduction, node pruning, labeling pruning, fast messages) (ablation) | Our-TRWS · Protein folding 1CKK | extra time for persistency: 268s -> 2.0s (persistency 26.53% unchanged) | Improved |
| add speedups (reduction through fast messages) (ablation) | Our-TRWS · colorseg-n4 pfau-small | extra time and persistency: 439s -> 19s; 88.59% -> 93.41% | Improved |
| add per-label persistency instead of whole-variable persistency | Our-CPLEX vs [43]-CPLEX · ProteinFolding 1SY9 | persistency: 0% -> 51% | Improved |
| use method with CPLEX instead of [43] on PDB1B25 | persistency method · ProteinFolding PDB1B25 | persistency: 22% -> 84% | Improved |
| replace exact CPLEX with suboptimal TRW-S dual solver | Our method · Random instances 20x20 Potts-3 | persistency / time: 87.38% -> 87.38%; 2.43s -> 0.06s | No clear effect |
| replace exact CPLEX with TRW-S on 20x20 Potts-4 | Our method · Random instances 20x20 Potts-4 | persistency / time: 74.28% -> 73.63%; 8.56s -> 0.08s | Mixed |
| use TRW-S instead of CPLEX for the baseline [43] | [43] · Random instances 10x10 full-3 | persistency: 2.64% -> 1.22% | Hurt |
| use proposed method instead of [43] on full-label random instances | persistency method · Random instances 20x20 full-3 | persistency: 0.83% -> 72.31% | Improved |
| apply method to OpenGM mrf-stereo vs MQPBO | Our-TRWS · OpenGM mrf-stereo | persistency: 0.23% -> 73.56% | Improved |
Closest papers by meaning
- Partial Optimality by Pruning for MAP-Inference with General Graphical Models (TPAMI 2016 · similarity 0.92)
- Exact MAP-Inference by Confining Combinatorial Search With LP Relaxation (AAAI 2018 · similarity 0.90)
- Inference in Graphical Models via Semidefinite Programming Hierarchies (NeurIPS 2017 · similarity 0.90)
- MAP Inference Via $$\ell _2$$-Sphere Linear Program Reformulation (IJCV 2020 · similarity 0.89)
- Exact MAP Inference by Avoiding Fractional Vertices (ICML 2017 · similarity 0.89)
- Anytime Best+Depth-First Search for Bounding Marginal MAP (AAAI 2017 · similarity 0.88)
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.
