Research map / Learning theory and algorithms / Clustering
Multi-Scale Fusion Subspace Clustering Using Similarity Constraint
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Fuse self-expression coefficient matrices from all encoder scales with a convolutional kernel and constrain the fused matrix with self-expression loss (ablation) | SC-MSFSC · ORL | clustering error: 14.00 -> 2.00 | Helped |
| Add the fused self-expression loss L3 (ablation) | SC-MSFSC · Extended Yale B, 38 subjects | clustering error: 4.85 -> 0.29 | Helped |
| Add the fused self-expression loss L3 (ablation) | SC-MSFSC · ORL | clustering error: 3.00 -> 3.75 | Hurt |
| Add the similarity constraint loss L4 (ablation) | SC-MSFSC · ORL | clustering error: 3.75 -> 2.00 | Helped |
| Add the similarity constraint loss L4 (ablation) | SC-MSFSC · Extended Yale B, 38 subjects | clustering error: 0.29 -> 0.29 | No clear effect |
| Use coefficient matrices from all three encoder layers instead of only layer 3 (ablation) | SC-MSFSC · ORL | clustering error: 26.75 -> 2.00 | Helped |
| Use coefficient matrices from all three encoder layers instead of layers 2 and 3 (ablation) | SC-MSFSC · ORL | clustering error: 12.25 -> 2.00 | Helped |
| Fuse layer-specific matrices with a 3 × 3 convolutional kernel instead of summing them (ablation) | SC-MSFSC · ORL | clustering error: 14.25 -> 2.00 | Helped |
| Use a 3 × 3 fusion kernel instead of a 1 × 1 fusion kernel (ablation) | SC-MSFSC · ORL | clustering error: 25.50 -> 2.00 | Helped |
| Use a 3 × 3 fusion kernel instead of a 5 × 5 fusion kernel (ablation) | SC-MSFSC · ORL | clustering error: 10.25 -> 2.00 | Helped |
| Initialize the fusion kernel with Glorot Uniform instead of Glorot Normal (ablation) | SC-MSFSC · ORL | clustering error: 11.50 -> 2.00 | Helped |
| Initialize the fusion kernel with Glorot Uniform instead of Random Normal (ablation) | SC-MSFSC · ORL | clustering error: 42.25 -> 2.00 | Helped |
Closest papers by meaning
- Exploring a Principled Framework for Deep Subspace Clustering (ICLR 2025 · similarity 0.91)
- Deep Subspace Clustering Networks (NeurIPS 2017 · similarity 0.91)
- LRSC: Learning Representations for Subspace Clustering (AAAI 2021 · similarity 0.91)
- Efficient Deep Embedded Subspace Clustering (CVPR 2022 · similarity 0.90)
- Customized Multiple Clustering via Multi-Modal Subspace Proxy Learning (NeurIPS 2024 · similarity 0.85)
- Contrastive Multi-view Subspace Clustering via Tensor Transformers Autoencoder (AAAI 2025 · similarity 0.85)
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.