Research map / Learning theory and algorithms / Clustering
Sparse Quantized Spectral Clustering
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 |
|---|---|---|---|
| Apply selective, data-dependent thresholding instead of uniform random sparsification | Spectral clustering kernel matrix · Spectral clustering | sparsity at matched clustering performance: ε_unif = erfc(s) + 2 s e^{-s^2}/ √π > erfc(s) = ε_selec | Helped |
| Increase the truncation threshold for selective sparsification (ablation) | Sparse operator f1 · Spectral clustering | classification error | Hurt |
| Use quantization or binarization instead of the dense linear kernel | Spectral clustering kernel matrix · Spectral clustering | storage reduction: 32 or 64 | Helped |
| Use optimally thresholded binary f3 rather than the sign function (ablation) | Binary operator f3 · Spectral clustering | nu/a1^2: π/2 ≈ 1.57 -> 1.24 | Helped |
| Choose the optimal threshold for quantized or binary operators (ablation) | Operators f2 and f3 · Spectral clustering | classification error | Helped |
| Use quantization f2 rather than sparsification f1 at approximately matched storage | Spectral clustering kernel matrix · Spectral clustering | classification error | Helped |
| Use binarization f3 rather than sparsification f1 at approximately matched storage | Spectral clustering kernel matrix · Spectral clustering | classification error | Helped |
| Allow a2 to be nonzero rather than setting a2=0 (ablation) | Nonlinear kernel transformation · Spectral clustering | clustering usefulness of the top eigenvector | Hurt |
| Use a nonlinear transform with nu/a1^2 greater than 1 instead of the linear function (ablation) | Spectral clustering kernel matrix · Spectral clustering | classification error | Hurt |
| Increase the truncation threshold for quantized or binary operators away from their optimum (ablation) | Operators f2 and f3 · Spectral clustering | classification error | Mixed |
| Use quantization f2 rather than binarization f3 | Spectral clustering kernel matrix · Spectral clustering | classification error | Mixed |
Closest papers by meaning
- Compressive Spectral Clustering (ICML 2016 · similarity 0.90)
- Nonconvex Sparse Spectral Clustering by Alternating Direction Method of Multipliers and Its Convergence Analysis (AAAI 2018 · similarity 0.89)
- A Tighter Analysis of Spectral Clustering, and Beyond (ICML 2022 · similarity 0.89)
- Structure-Aware Spectral Sparsification via Uniform Edge Sampling (NeurIPS 2025 · similarity 0.88)
- Pass-Efficient Algorithms for Graph Spectral Clustering (Student Abstract) (AAAI 2024 · similarity 0.88)
- Nonlinear Laplacians: Tunable principal component analysis under directional prior information (NeurIPS 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.