Research map / Learning theory and algorithms / Matrix and tensor methods
Dictionary Learning for Massive Matrix Factorization
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 |
|---|---|---|---|
| random subsampling of rows with mask ratio r (about 10%) in online dictionary learning (ablation) | sparse matrix factorization on HCP fMRI · HCP fMRI (2TB) | convergence time speed-up: near x r speed-up; ~10x | Helped |
| masked online dictionary learning (MODL) instead of coordinate descent | collaborative filtering · Netflix (140M) | convergence time: 6.8x faster (256 s vs 1714 s) | Helped |
| masked online dictionary learning instead of coordinate descent | collaborative filtering · MovieLens 1M | convergence time: 8 s vs 6 s (x0.75) | Hurt |
| masked online dictionary learning instead of coordinate descent | collaborative filtering · MovieLens 10M | convergence time: 3.7x faster (60 s vs 223 s) | Helped |
| masked online dictionary learning instead of coordinate descent | collaborative filtering · MovieLens 10M and Netflix | test RMSE: 0.802 -> 0.799; 0.938 -> 0.934 | Helped |
| approximate low-dimension projection instead of exact lazy projection on the dictionary (ablation) | collaborative filtering (MODL variant) · collaborative filtering | test RMSE | No clear effect |
| sampling step size exponent beta=0.9 for weights 1/t^beta (ablation) | MODL on collaborative filtering · MovieLens and Netflix | convergence speed | Mixed |
| subsampling ratio r up to 12 on fMRI with high regularization (ablation) | sparse dictionary learning on HCP fMRI · HCP fMRI | l1/l2 sparsity ratio: about 5% lower l1/l2 at r=12 for high regularization | Mixed |
Closest papers by meaning
- Compressed Factorization: Fast and Accurate Low-Rank Factorization of Compressively-Sensed Data (ICML 2019 · similarity 0.88)
- Scalable Probabilistic Matrix Factorization with Graph-Based Priors (AAAI 2020 · similarity 0.87)
- Learning brain regions via large-scale online structured sparse dictionary learning (NeurIPS 2016 · similarity 0.87)
- Connectivity Shapes Implicit Regularization in Matrix Factorization Models for Matrix Completion (NeurIPS 2024 · similarity 0.85)
- Two Sparse Matrices are Better than One: Sparsifying Neural Networks with Double Sparse Factorization (ICLR 2025 · similarity 0.85)
- In-depth Analysis of Low-rank Matrix Factorisation in a Federated Setting (AAAI 2025 · similarity 0.85)
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.