Research map / Learning theory and algorithms / Matrix and tensor methods
Coresets for Archetypal Analysis
Sebastian Mair, Ulf Brefeld
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 abs-cs instead of uniform sampling | archetypal analysis coreset · Covertype, k=25, m=1000 | relative error (%): 181.7% -> 148.9% | Improved |
| use abs-cs instead of lw-cs | archetypal analysis coreset · Song, k=25, m=5000 | relative error (%): 17.7% -> 12.4% | Improved |
| use abs-cs instead of lw-cs | archetypal analysis coreset · Pose, k=25, m=1000 | relative error (%): 14.2% -> 15.7% | Hurt |
| use abs-cs instead of lw-cs | archetypal analysis coreset · Ijcnn1, k=25, m=1000 | relative error (%): 8.9% -> 8.5% | No clear effect |
| use abs-cs instead of lw-cs | archetypal analysis coreset · Pose, k=25, m=5000 | relative error (%): 5.6% -> 5.5% | No clear effect |
| increase coreset size from m=1000 to m=5000 | abs-cs · Covertype, k=25 | relative error (%): 148.9% -> 79.1% | Improved |
| increase coreset size from m=1000 to m=5000 | abs-cs · Covertype, k=25 | speedup: 601x -> 111x | Hurt |
| use abs-cs instead of lucic-cs | archetypal analysis coreset · Song, k=25, m=5000 | relative error (%): 17.5% -> 12.4% | Improved |
| use abs-cs instead of lucic-cs | archetypal analysis coreset · Ijcnn1, k=25, m=5000 | relative error (%): 5.1% -> 4.0% | Improved |
| use abs-cs instead of uniform sampling | archetypal analysis coreset · Ijcnn1, k=25, m=5000 | relative error (%): 4.5% -> 4.0% | No clear effect |
| compute archetypes on the frame (convex hull boundary points) instead of a coreset | frame · Ijcnn1, k=100 | relative error: 0.03 | Improved |
Closest papers by meaning
- Universal Weak Coreset (AAAI 2024 · similarity 0.82)
- Beyond Worst-Case Coreset Bounds for $k$-Clustering via Determinantal Sampling (NeurIPS 2026 · similarity 0.82)
- Simple Weak Coresets for Non-decomposable Classification Measures (AAAI 2024 · similarity 0.81)
- Archetypal SAE: Adaptive and Stable Dictionary Learning for Concept Extraction in Large Vision Models (ICML 2025 · similarity 0.81)
- Robust Sparsification via Sensitivity (ICML 2025 · similarity 0.80)
- Coresets for Multiple $\ell_p$ Regression (ICML 2024 · similarity 0.80)
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.
