Research map / Learning theory and algorithms / Matrix and tensor methods
Frame-based Data Factorizations
Sebastian Mair, Ahcène Boubekki, 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 |
|---|---|---|---|
| restrict archetype optimization to the frame (Frame-AA vs AA) | Archetypal Analysis · USAFSurvey, p=6 | Frobenius error: 904.22 vs 902.07 | No clear effect |
| Frame-AA vs AA on ozone | Archetypal Analysis · ozone, p=6 | Frobenius error: 1669.70 -> 1532.12 | Improved |
| Frame-AA vs AA on Banking2 | Archetypal Analysis · Banking2, p=6 | Frobenius error: 72.35 -> 90.08 | Hurt |
| Frame-AA vs CH-NMF | CH-NMF · USAFSurvey, p=6 | Frobenius error: 1239.67 -> 904.22 | Improved |
| Frame-AA vs Convex-NMF | C-NMF · USAFSurvey, p=6 | Frobenius error: 2192.30 -> 904.22 | Improved |
| use standard NMF instead of Frame-AA | Frame-AA · USAFSurvey, p=6 | Frobenius error: 904.22 -> 688.35 | Improved |
| NNLS-Frame discovers up to d+1 extreme points per iteration vs LP-based one per iteration | frame computation · synthetic n=2500, d=5 | extreme points discovered vs iterations | Improved |
| NNLS-Frame vs LP-based frame methods | frame computation · synthetic n=10000 | runtime | Improved |
| divide and conquer with three partitions | NNLS-Frame · synthetic n=10000, d=5 | runtime | Improved |
| reuse frame across multiple p values | Frame-AA · USAFSurvey, p=4..16 | cumulative time | Improved |
| Frame-AA embedding vs convolutional autoencoder embedding with SVM | Frame-AA · MNIST 1 vs 7 | classification accuracy | Improved |
| Frame-AA vs convolutional autoencoder | Frame-AA · MNIST 1 vs 7 | reconstruction error | Hurt |
Closest papers by meaning
- Coresets for Archetypal Analysis (NeurIPS 2019 · similarity 0.90)
- Online Dictionary Learning for Approximate Archetypal Analysis (ECCV 2018 · similarity 0.86)
- A Geometric Approach to Archetypal Analysis via Sparse Projections (ICML 2020 · similarity 0.86)
- Sparse NMF with Archetypal Regularization: Computational and Robustness Properties (JMLR 2024 · similarity 0.85)
- Generalized Leverage Scores: Geometric Interpretation and Applications (ICML 2022 · similarity 0.84)
- Equivariant Frames and the Impossibility of Continuous Canonicalization (ICML 2024 · similarity 0.82)
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.
