Research map / Optimization / Nonconvex and smooth optimization
Convex Relaxations for Manifold-Valued Markov Random Fields with Approximation Guarantees
Robin Kenis, Emanuel Laude, Panagiotis Patrinos
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 the polynomial SDP relaxation instead of discretized linear (label-sampling) relaxations | geodesic chromaticity denoising on S2 · Rooftop/Flowers/Butterfly images | primal energy at 1000 s: Rooftop P 3228.25 vs 3288.05 (TRW-S 150 labels) | Improved |
| use linear label-discretization relaxation with up to 150 labels | TRW-S on chromaticity denoising · Butterfly image | convergence to optimum | Hurt |
| use the SDP relaxation on an adversarial S1-valued toy problem with a 180 degree jump | SDP relaxation vs discrete relaxation · S1-valued toy dataset | cost at equal memory | Hurt |
| increase LR-Solver labels from 150 to 200 (ablation) | LR-Solver · Rooftop denoising 1000 s | primal energy and dual bound: P 3287.44 -> 3402.78; D 3287.44 -> 2782.90 | Hurt |
| use degree-2 relaxation with least-squares or ManOpt rounding | pose graph optimization · INTEL, MIT, M3500b | qualitative solution | Mixed |
Closest papers by meaning
- Continuous Relaxation of MAP Inference: A Nonconvex Perspective (CVPR 2018 · similarity 0.88)
- Sublabel-Accurate Relaxation of Nonconvex Energies (CVPR 2016 · similarity 0.88)
- Convex Relaxations of ReLU Neural Networks Approximate Global Optima in Polynomial Time (ICML 2024 · similarity 0.87)
- Weakly Convex Regularisers for Inverse Problems: Convergence of Critical Points and Primal-Dual Optimisation (ICML 2024 · similarity 0.85)
- Spectral Preconditioning for Gradient Methods on Graded Non-convex Functions (ICML 2024 · similarity 0.85)
- Convex Formulations for Training Two-Layer ReLU Neural Networks (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.
