Research map / Learning theory and algorithms / Combinatorial and search optimization
Biconvex Relaxation for Semidefinite Programming in Computer Vision
Sohil Shah, Abhay Kumar Yadav, Carlos D. Castillo, David W. Jacobs, Christoph Studer, Tom Goldstein
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 |
|---|---|---|---|
| use BCR instead of SDCut | image segmentation solver · Berkeley image segmentation | time (s): 27.64 -> 0.97 | Improved |
| use BCR instead of SDCut | image segmentation solver · Berkeley image segmentation | objective: 6.40 -> 6.34 | Improved |
| use BCR instead of BNCut (rank 2 vs rank 1) | image segmentation solver · Berkeley image segmentation | objective: 10.84 -> 6.34 | Improved |
| use BNCut instead of BCR | image segmentation solver · Berkeley image segmentation | time (s): 0.97 -> 0.08 | Improved |
| use BCR instead of SDCut | co-segmentation solver · Co-segmentation (horse) | time (s): 220 -> 18.8 | Improved |
| use BCR instead of SDCut | co-segmentation solver · Co-segmentation (face) | objective: 4.94 -> 3.29 | Improved |
| use BCR instead of low-rank factorization | co-segmentation solver · Co-segmentation (horse) | time (s): 2647 -> 18.8 | Improved |
| use BCR instead of low-rank factorization | co-segmentation solver · Co-segmentation (car-back) | objective: 5.00 -> 4.36 | Hurt |
| restrict BCRML to rank 100 instead of full dimension (ablation) | BCRML · YTC | training time (s): 189 -> 45 | Improved |
| restrict BCRML to rank 100 instead of full dimension (ablation) | BCRML · YTF | accuracy (%): 60.58 -> 62.42 | Improved |
| use BCRML-100 instead of SPDML-Stein | metric learning · YTC | accuracy (%): 66.10 -> 64.61 | Hurt |
| add BCRML-100 before CDL-LDA (ablation) | CDL-LDA · YTC | total time (s): 302.7 -> 57.7 | Improved |
Closest papers by meaning
- Large-Scale Binary Quadratic Optimization Using Semidefinite Relaxation and Applications (TPAMI 2016 · similarity 0.90)
- Deterministic Consensus Maximization with Biconvex Programming (ECCV 2018 · similarity 0.87)
- Efficient semidefinite-programming-based inference for binary and multi-class MRFs (NeurIPS 2020 · similarity 0.86)
- Convex Relaxations for Manifold-Valued Markov Random Fields with Approximation Guarantees (ECCV 2024 · similarity 0.86)
- Sublabel-Accurate Relaxation of Nonconvex Energies (CVPR 2016 · similarity 0.85)
- Moreau Envelope for Nonconvex Bi-Level Optimization: A Single-Loop and Hessian-Free Solution Strategy (ICML 2024 · 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.
