Research map / Learning theory and algorithms / Clustering
Deep Spectral Clustering Learning
Read the paperCheck 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 |
|---|---|---|---|
| add spectral clustering (Algorithm 2) on top of learned DSCL features instead of k-means on l2-normalized features (ablation) | DSCL last layer · CUB-200-2011 (Birds) | R@1: 50.08 -> 53.22 | Helped |
| add spectral clustering on top of learned DSCL features instead of k-means (ablation) | DSCL last layer · Cars196 | R@1: 67.54 -> 73.07 | Helped |
| add spectral clustering on top of learned DSCL features instead of k-means (ablation) | DSCL end-to-end · Products | R@1: 64.52 -> 67.59 | Helped |
| end-to-end fine-tuning instead of last-layer-only fine-tuning (ablation) | DSCL · Products | R@1: 65.30 -> 67.59 (with SC) | Helped |
| last-layer-only fine-tuning instead of end-to-end fine-tuning (ablation) | DSCL · Birds and Cars196 | R@1: Birds 49.78 -> 53.22; Cars 59.37 -> 73.07 (with SC) | Helped |
| DSCL with spectral clustering vs NMI-based deep clustering baseline | Birds, Cars196, Products · CUB-200-2011 / Cars196 / Products | Recall@K: Birds R@1 48.18 -> 53.22; Cars R@1 58.11 -> 73.07; Products R@1 67.02 -> 67.59 | Helped |
| DSCL with spectral clustering vs NMI-based baseline on NMI | Birds and Products · CUB-200-2011 / Products | NMI: 59.16 vs 59.23 (Birds); 89.40 vs 89.48 (Products) | Hurt |
| learned DSCL features with plain k-means instead of spectral clustering on test set (ablation) | DSCL on Products · Products | R@1: 67.59 -> 64.52 (end-to-end) | Hurt |
Closest papers by meaning
- SpectralNet: Spectral Clustering using Deep Neural Networks (ICLR 2018 · similarity 0.91)
- Deep Spectral Clustering Using Dual Autoencoder Network (CVPR 2019 · similarity 0.90)
- Clustering-friendly Representation Learning via Instance Discrimination and Feature Decorrelation (ICLR 2021 · similarity 0.89)
- Mini-cluster Guided Long-tailed Deep Clustering (ICLR 2026 · similarity 0.88)
- Unified Spectral Clustering With Optimal Graph (AAAI 2018 · similarity 0.88)
- From Logits to Hierarchies: Hierarchical Clustering made Simple (ICML 2025 · similarity 0.86)
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.