Research map / Graph learning / Graph algorithms and clustering
End-to-End Learning for Graph Decomposition
Jie Song, Bjoern Andres, Michael J. Black, Otmar Hilliges, Siyu Tang
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 jointly trained CRF high-order cycle potentials to Siamese network | Siamese network clustering · MNIST clustering | clustering accuracy: 94.1% -> 95.9% | Improved |
| add jointly trained CRF high-order cycle potentials to Siamese network | Siamese network · MNIST similarity | similarity accuracy: 91.5% -> 93.2% | Improved |
| add end-to-end finetuning over unary-and-CRF (ablation) | multicut pose association (greedy) · MPII validation | mean AP: 73.78 -> 74.36 | Improved |
| add end-to-end finetuning over unary-and-CRF (ablation) | multicut pose association (KL) · MPII validation | mean AP: 71.96 -> 72.57 | Improved |
| add CRF on top of unary (offline) (ablation) | multicut pose association (greedy) · MPII validation | mean AP: 73.21 -> 73.78 | Improved |
| add CRF on top of unary (offline) (ablation) | multicut pose association (KL) · MPII validation | mean AP: 71.32 -> 71.96 | Improved |
| use greedy heuristic instead of KL heuristic for feasible solution (ablation) | end-to-end model · MPII validation | mean AP: 72.57 -> 74.36 | Improved |
| increase mean-field iterations from 0 to 3 (ablation) | CRF inference · MPII limb links | mean marginal probability: 0.635 -> 0.674 | Improved |
| increase mean-field iterations from 2 to 3 (ablation) | CRF inference · MPII limb links | mean marginal probability: 0.671 -> 0.674 | No clear effect |
| increase mean-field iterations from 0 to 3 (ablation) | CRF inference · MPII cliques | ratio of non-valid cycles (%): 2.60 -> 1.76 | Improved |
Closest papers by meaning
- Joint Graph Decomposition & Node Labeling: Problem, Algorithms, Applications (CVPR 2017 · similarity 0.87)
- OPTFM: A Scalable Multi-View Graph Transformer for Hierarchical Pre-Training in Combinatorial Optimization (NeurIPS 2025 · similarity 0.85)
- Graph Positional Encoding via Random Feature Propagation (ICML 2023 · similarity 0.84)
- Efficient Graph Cut Optimization for Full CRFs with Quantized Edges (TPAMI 2019 · similarity 0.84)
- Permutation-Invariant Variational Autoencoder for Graph-Level Representation Learning (NeurIPS 2021 · similarity 0.84)
- Discriminative Training of Conditional Random Fields with Probably Submodular Constraints (IJCV 2020 · 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.
