Research map / Learning theory and algorithms / Matrix and tensor methods
Accelerating SGD for Highly Ill-Conditioned Huge-Scale Online Matrix Completion
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 |
|---|---|---|---|
| Precondition SGD updates with (XᵀX)⁻¹ | SGD for symmetric matrix completion · RMSE loss, local convergence | iterations to ε-accuracy: O(κ⁴·dr log(d/ε)) -> O(dr log(d/ε)) | Helped |
| Precondition SGD updates with (XᵀX)⁻¹ | SGD for item-item collaborative filtering · MovieLens25M, BPR loss | training samples to reach NP-Maximum AUC: 46% -> 11% | Helped |
| Precondition SGD updates with (XᵀX)⁻¹ | SGD for item-item collaborative filtering · MovieLens25M, BPR loss | training samples to reach 0.9 AUC: 81% -> 16% | Helped |
| Precondition SGD updates with (XᵀX)⁻¹ | SGD for matrix completion · Synthetic RMSE, ill-conditioned ground truth (κ(M) = 10⁴) | convergence rate | Helped |
| Precondition SGD updates with (XᵀX)⁻¹ | SGD for Euclidean distance matrix completion · EDM completion with outlier points | convergence rate | Helped |
| Precondition SGD updates with (XᵀX)⁻¹ | SGD for matrix completion · Noisy RMSE and pointwise cross-entropy experiments | convergence toward noise floor | Helped |
Closest papers by meaning
- Provable Efficient Online Matrix Completion via Non-convex Stochastic Gradient Descent (NeurIPS 2016 · similarity 0.93)
- Rank-1 Matrix Completion with Gradient Descent and Small Random Initialization (NeurIPS 2023 · similarity 0.91)
- Symmetric Matrix Completion with ReLU Sampling (ICML 2024 · similarity 0.90)
- A Scalable Second Order Method for Ill-Conditioned Matrix Completion from Few Samples (ICML 2021 · similarity 0.89)
- Generalization Analysis of Deep Non-linear Matrix Completion (ICML 2024 · similarity 0.87)
- RGNMR: A Gauss-Newton method for robust matrix completion with theoretical guarantees (NeurIPS 2025 · 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.