Research map / Learning theory and algorithms / Matrix and tensor methods
Alternating minimization for dictionary learning with random initialization
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 |
|---|---|---|---|
| Replace operator-norm bound on the true dictionary A* with an infinity-norm (max-entry) bound ||A*||_inf <= C_b/s | alternating minimization for dictionary learning · theoretical analysis, sparse dictionary learning | allowed dictionary scale / convergence guarantee | Helped |
| Restrict sparsity to s = O(sqrt(d)) instead of s = O(d^{1/6}) (Agarwal et al.) or s = O(sqrt(d)/log d) (Arora et al.) | alternating minimization with l1 MU-selector and gradient step · overcomplete dictionary learning, theory | maximum tolerable sparsity level | Helped |
| Use {l1,l2,linf}-MU Selector sparse estimator instead of standard LASSO/Dantzig selector in the sparse-coding step | alternating minimization for dictionary learning · theory, dictionary corruption | radius of convergence / error guarantee | Helped |
| Random initialization replaced by requiring initial dictionary within inf-norm C_b/(2s) of A* (Assumption B1), after an erratum | alternating minimization with random initialization · theory, erratum | convergence guarantee | No clear effect |
Closest papers by meaning
- NOODL: Provable Online Dictionary Learning and Sparse Coding (ICLR 2019 · similarity 0.90)
- Understanding l4-based Dictionary Learning: Interpretation, Stability, and Robustness (ICLR 2020 · similarity 0.89)
- Understanding approximate and unrolled dictionary learning for pattern recovery (ICLR 2022 · similarity 0.89)
- On Learning Sparsely Used Dictionaries from Incomplete Samples (ICML 2018 · similarity 0.89)
- The Generalization Error of Dictionary Learning with Moreau Envelopes (ICML 2018 · similarity 0.88)
- On the Sparsity-Storage-Accuracy Tradeoff in Parsimoniously Activated Dictionary Learning (NeurIPS 2026 · similarity 0.86)
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.