Research map / Learning theory and algorithms / Clustering
SpectralNet: Spectral Clustering using Deep Neural Networks
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 |
|---|---|---|---|
| replace Euclidean affinity with Siamese distance (ablation) | SpectralNet in MNIST input space · MNIST | ACC: .622$\pm$.008 -> .826$\pm$.03 | Helped |
| replace Euclidean affinity with Siamese distance (ablation) | SpectralNet in MNIST input space · MNIST | NMI: .687$\pm$.004 -> .884$\pm$.02 | Helped |
| replace Euclidean affinity with Siamese distance (ablation) | SpectralNet in MNIST code space · MNIST | ACC: .800$\pm$.003 -> .971$\pm$.001 | Helped |
| replace Euclidean affinity with Siamese distance (ablation) | SpectralNet in MNIST code space · MNIST | NMI: .814$\pm$.008 -> .924$\pm$.001 | Helped |
| replace Euclidean affinity with Siamese distance (ablation) | SpectralNet in Reuters input space · Reuters | ACC: .645$\pm$.01 -> .661$\pm$ 017 | Helped |
| replace Euclidean affinity with Siamese distance (ablation) | SpectralNet in Reuters input space · Reuters | NMI: .444$\pm$.01 -> .381 $\pm$ .057 | Hurt |
| replace Euclidean affinity with Siamese distance (ablation) | SpectralNet in Reuters code space · Reuters | ACC: .605$\pm$.053 -> .803$\pm$.006 | Helped |
| replace Euclidean affinity with Siamese distance (ablation) | SpectralNet in Reuters code space · Reuters | NMI: .401$\pm$.061 -> .532$\pm$.010 | Helped |
| use autoencoder code space instead of pixel input space, with Euclidean affinity (ablation) | SpectralNet · MNIST | ACC: .622$\pm$.008 -> .800$\pm$.003 | Helped |
| use autoencoder code space instead of pixel input space, with Siamese affinity (ablation) | SpectralNet · MNIST | ACC: .826$\pm$.03 -> .971$\pm$.001 | Helped |
| use autoencoder code space instead of input space, with Siamese affinity (ablation) | SpectralNet · Reuters | ACC: .661$\pm$ 017 -> .803$\pm$.006 | Helped |
| use autoencoder code space instead of input space, with Siamese affinity (ablation) | SpectralNet · Reuters | NMI: .381 $\pm$ .057 -> .532$\pm$.010 | Helped |
Closest papers by meaning
- Deep Spectral Clustering Using Dual Autoencoder Network (CVPR 2019 · similarity 0.93)
- Deep Spectral Clustering Learning (ICML 2017 · similarity 0.91)
- Spectral Clustering with Graph Neural Networks for Graph Pooling (ICML 2020 · similarity 0.90)
- Delving into Spectral Clustering with Vision-Language Representations (ICLR 2026 · similarity 0.90)
- Fast and Simple Spectral Clustering in Theory and Practice (NeurIPS 2023 · similarity 0.90)
- HoloNets: Spectral Convolutions do extend to Directed Graphs (ICLR 2024 · 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.