Research map / Learning theory and algorithms / Matrix and tensor methods
SPALS: Fast Alternating Least Squares via Implicit Leverage Scores Sampling
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 |
|---|---|---|---|
| increase sampling rate alpha from 0.3 to 3.0 (more sampled rows per iteration) (ablation) | SPALS · Amazon review tensor | relative error: 0.987 -> 0.982 (small gain), time 6.97s -> 38.9s | Mixed |
| SPALS sampling (alpha=1) vs. ALS-dense and sketch baselines on dense synthetic tensor | SPALS · Dense synthetic 1000^3 nsr=0.1 | relative error: 0.18 vs 0.27 (ALS-dense), 0.45 (sketch) | Helped |
| SPALS sampling vs. sketch-based CP baseline at similar cost | SPALS · Dense synthetic 1000^3 nsr=1 and 10 | relative error: 1.10 vs 1.37 (nsr=1); 10.21 vs 11.11 (nsr=10) | No clear effect |
| SPALS sampling vs. deterministic ALS on Amazon tensor | SPALS · Amazon review tensor | time per iteration: 15.7s vs 142s per iteration (SPALS(1) vs ALS-sparse) | Helped |
| SPALS sampling vs. deterministic ALS on Amazon tensor (error) | SPALS · Amazon review tensor | relative error: 0.983 vs 0.981 | No clear effect |
| sampling rate alpha from 0.3 to 3.0 on dense synthetic tensors (time) (ablation) | SPALS · Dense synthetic 1000^3 nsr=0.1 | time per iteration: 1.76s (0.3) vs 15.9s (3.0) | Hurt |
Closest papers by meaning
- More Efficient Sampling for Tensor Decomposition With Worst-Case Guarantees (ICML 2022 · similarity 0.92)
- Efficient Leverage Score Sampling for Tensor Train Decomposition (NeurIPS 2024 · similarity 0.90)
- Fused Orthogonal Alternating Least Squares for Tensor Clustering (NeurIPS 2022 · similarity 0.90)
- A Sampling-Based Method for Tensor Ring Decomposition (ICML 2021 · similarity 0.90)
- Fast Tensor Completion via Approximate Richardson Iteration (ICML 2025 · similarity 0.88)
- Towards Multi-Mode Outlier Robust Tensor Ring Decomposition (AAAI 2024 · similarity 0.87)
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.