Research map / Learning theory and algorithms / Clustering
Subspace Clustering by Block Diagonal Representation
Canyi Lu, Jiashi Feng, Zhouchen Lin, Tao Mei, Shuicheng Yan
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 |
|---|---|---|---|
| replace SSC sparsity prior with block diagonal regularizer (BDR-Z) | SSC · Hopkins 155 (2F-dim) | clustering error mean (%): 2.18 -> 0.93 | Improved |
| use block diagonal regularizer (BDR-Z) over best baseline S3C | S3C · Hopkins 155 (4k-dim PCA) | clustering error mean (%): 2.41 -> 1.08 | Improved |
| project data to 4k-dim with PCA instead of 2F-dim | BDR-Z · Hopkins 155 | clustering error mean (%): 0.93 -> 1.08 | Hurt |
| compare BDR-Z against LatLRR with its heavy pre/post-processing | BDR-Z · Hopkins 155 (reported settings) | clustering error mean (%): 0.93 -> 0.85 | Hurt |
| add block diagonal regularizer (BDR-B) vs SSC with more subjects | SSC · Extended Yale B (10 subjects) | clustering error mean (%): 10.09 -> 2.95 | Improved |
| add block diagonal regularizer (BDR-B) vs SSC with few subjects | SSC · Extended Yale B (2 subjects) | clustering error mean (%): 1.64 -> 3.28 | Hurt |
| use Z instead of B as affinity (BDR-Z vs BDR-B) (ablation) | BDR · Extended Yale B (3 subjects) | clustering error mean (%): 3.02 -> 1.15 | Mixed |
| use Z instead of B as affinity (BDR-Z vs BDR-B) (ablation) | BDR · Hopkins 155 (2F-dim) | clustering error mean (%): 1.22 -> 0.93 | No clear effect |
| use BDR-B over LSR as number of subjects grows | LSR · Extended Yale B (10 subjects) | clustering error mean (%): 33.27 -> 2.95 | Improved |
| set lambda and gamma too small or too large (ablation) | BDR · Extended Yale B (10 subjects) | clustering error | Hurt |
Closest papers by meaning
- Maximum Block Energy Guided Robust Subspace Clustering (TPAMI 2022 · similarity 0.90)
- Stochastic Sparse Subspace Clustering (CVPR 2020 · similarity 0.90)
- Sparse Subspace Clustering by Learning Approximation ℓ0 Codes (AAAI 2017 · similarity 0.89)
- Hyperplane Clustering via Dual Principal Component Pursuit (ICML 2017 · similarity 0.89)
- Subspace Clustering With Priors via Sparse Quadratically Constrained Quadratic Programming (CVPR 2016 · similarity 0.88)
- Efficient Sparse PCA via Block-Diagonalization (ICLR 2025 · 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.
