Research map / Learning theory and algorithms / Clustering
Recursive Nearest Agglomeration (ReNA): Fast Clustering for Approximation of Structured Signals
Andrés Hoyos-Idrobo, Gael P. Varoquaux, Jonas Kahn, Bertrand Thirion
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 |
|---|---|---|---|
| use balanced-cluster feature grouping (Ward, SLIC, ReNA) instead of random projections or downsampling at high reduction | logistic regression · Extended Yale B faces | accuracy: up to 10% better | Improved |
| use ReNA feature grouping before classifier instead of raw data | logistic regression · OASIS, HCP and other datasets | prediction accuracy: p-value < 10^-4 | Improved |
| use ReNA instead of raw data | l2 logistic regression · Extended Yale B faces | accuracy: slightly worse than raw | Hurt |
| use percolating linkage (single, average, complete) for feature grouping | logistic regression · Extended Yale B faces | accuracy: fail to achieve the same performance as raw | Hurt |
| add ICA after ReNA reduction instead of random projections or Nystrom | spatial ICA · HCP 40 components | average correlation with raw components: greater than 0.74 vs < 0.4 | Improved |
| use ReNA reduction before ICA | spatial ICA · HCP 93 subjects | computation time: 16 times faster than raw | Improved |
| use ReNA reduction before ICA | spatial ICA · HCP 93 subjects | total analysis time: factor of 15 | Improved |
| use Ward instead of ReNA for approximation | feature grouping · k = p/20 | approximation fidelity and time: slightly better approximations but slower | Mixed |
| downsample images before ICA instead of raw data | spatial ICA · HCP across two sessions | component consistency: more consistent than raw data | Improved |
Closest papers by meaning
- Robust Discriminative Clustering with Sparse Regularizers (JMLR 2017 · similarity 0.86)
- Simple, Scalable and Effective Clustering via One-Dimensional Projections (NeurIPS 2023 · similarity 0.85)
- Hyperplane Clustering via Dual Principal Component Pursuit (ICML 2017 · similarity 0.85)
- Achieving Optimal Clustering in Gaussian Mixture Models with Anisotropic Covariance Structures (NeurIPS 2024 · similarity 0.84)
- Capturing the denoising effect of PCA via compression ratio (NeurIPS 2024 · similarity 0.84)
- Near-Linear Time Approximation Algorithms for k-means with Outliers (ICML 2024 · 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.
