Research map / Optimization / Nonconvex and smooth optimization
Large-Scale Binary Quadratic Optimization Using Semidefinite Relaxation and Applications
Peng Wang, Chunhua Shen, Anton van den Hengel, Philip H. S. Torr
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 |
|---|---|---|---|
| increase regularization parameter gamma (ablation) | SDCut · graph bisection, random dense graph n=1000 | bound tightness and iterations | Mixed |
| use smoothing Newton instead of quasi-Newton | SDCut · dense graph bisection n=200 | iterations: 67.7 -> 11.0 | Improved |
| use smoothing Newton instead of quasi-Newton | SDCut · dense graph bisection n=1000 | time: 22.6s -> 13.0s | Improved |
| use quasi-Newton instead of smoothing Newton | SDCut · sparse graph bisection n=5000 | time: 26m19s -> 04m43s | Improved |
| run eigen-decomposition on CPU+GPU instead of one CPU core | SDCut-QN · dense graph bisection n=10000 | time: 3h8m -> 18m | Improved |
| use SDCut instead of interior-point SDP solver | SDCut-QN · segmentation with partial grouping constraints | time: 6m12s -> 23.7s | Improved |
| use SDP relaxation instead of spectral NCut | SDCut-QN · dense graph bisection n=1000 | upper-bound: 6.10 -> 5.06 | Improved |
| use SDCut with quadratic histogram constraint instead of graph cuts | SDCut-QN · segmentation with histogram constraints | F-measure: 0.722 -> 0.930 | Improved |
| use SDCut instead of constrained spectral SMQC | SDCut-QN · segmentation with histogram constraints | F-measure: 0.832 -> 0.930 | Improved |
| use SDCut-QN instead of SDCut-SN exploiting structure | SDCut-QN · co-segmentation face | time: 43m57s -> 8m18s | Improved |
| use SDCut-QN instead of low-rank factorization | SDCut-QN · co-segmentation face | upper-bound: 20.53 -> 12.65 | Improved |
| use SDCut instead of spectral SMAC | SDCut · graph matching 15x30 | error ratio: 6/150 -> 1/150 | Improved |
Closest papers by meaning
- Biconvex Relaxation for Semidefinite Programming in Computer Vision (ECCV 2016 · similarity 0.90)
- A Fast Binary Quadratic Programming Solver Based on Stochastic Neighborhood Search (TPAMI 2020 · similarity 0.89)
- The non-convex Burer-Monteiro approach works on smooth semidefinite programs (NeurIPS 2016 · similarity 0.87)
- BayeSQP: Bayesian Optimization through Sequential Quadratic Programming (NeurIPS 2025 · similarity 0.84)
- Efficient Quadratic Corrections for Frank-Wolfe Algorithms (NeurIPS 2025 · similarity 0.84)
- Deep Distributed Optimization for Large-Scale Quadratic Programming (ICLR 2025 · similarity 0.84)
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.
