Research map / Learning theory and algorithms / Clustering
Variable Clustering via Distributionally Robust Nodewise Regression
Kaizheng Wang, Xiao Xu, XUNYU ZHOU
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 Lasso nodewise regression with DRO nodewise regression | subspace clustering · Synthetic multi-factor block model (K=25, d=500, n=250) | AMI: 0.83 -> 0.92 | Improved |
| replace Lasso with DRO | subspace clustering · Extended Yale B (23 trials) | mean AMI: 0.403 -> 0.580 | Improved |
| replace Lasso with DRO | subspace clustering · Extended Yale B (23 trials) | median AMI: 0.422 -> 0.584 | Improved |
| replace MFC with DRO | variable clustering · Synthetic multi-factor block model (K=25, d=500, n=250) | AMI: 0.43 -> 0.92 | Improved |
| replace SSC with DRO | subspace clustering · Synthetic multi-factor block model (K=25, d=500, n=250) | AMI: 0.025 -> 0.92 | Improved |
| use DRO instead of correlation-based ACC | variable clustering · Synthetic multi-factor block model (K=25, d=500, n=250) | AMI: 0.15 -> 0.92 | Improved |
| solve DRO with ADMM instead of off-the-shelf convex optimizers | DRO clustering · Synthetic multi-factor block model (K=25, d=500, n=250) | runtime: over 80% | Improved |
| vary initial ADMM penalty rho0 from 0.01 to 10 (ablation) | DRO · Synthetic multi-factor block model (K=25, d=500, n=250) | AMI: 0.91 to 0.92 | No clear effect |
| overestimate cluster number slightly (K=27 vs 25) (ablation) | DRO · Synthetic multi-factor block model (K=25, d=500, n=250) | AMI: 0.92 | No clear effect |
| underestimate cluster number K (ablation) | DRO · Synthetic multi-factor block model (K=25, d=500, n=250) | AMI | Hurt |
| vary confidence level alpha from 0.001 to 0.2 (ablation) | DRO · Synthetic multi-factor block model (K=25, d=500, n=250) | AMI: 0.91 to 0.93 | No clear effect |
| increase global factor magnitude beta_H^2 from 0.1 to 0.9 | all clustering methods · Synthetic, noise var 1 | AMI | Hurt |
Closest papers by meaning
- Robust Subspace Clustering With Independent and Piecewise Identically Distributed Noise Modeling (CVPR 2019 · similarity 0.88)
- Robust Discriminative Clustering with Sparse Regularizers (JMLR 2017 · similarity 0.87)
- Decentralized Robust Subspace Clustering (AAAI 2016 · similarity 0.87)
- Subspace Structure-Aware Spectral Clustering for Robust Subspace Clustering (ICCV 2019 · similarity 0.87)
- Subspace Clustering With Priors via Sparse Quadratically Constrained Quadratic Programming (CVPR 2016 · similarity 0.87)
- Heterogeneous Sufficient Dimension Reduction and Subspace Clustering (ICML 2025 · similarity 0.84)
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.
