Research map / Learning theory and algorithms / Matrix and tensor methods
Using Side Information to Reliably Learn Low-Rank Matrices from Missing and Corrupted Observations
Kai-Yang Chiang, Inderjit S. Dhillon, Cho-Jui Hsieh
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 |
|---|---|---|---|
| use noise-tolerant IMCNF instead of IMC with noisy user features | Epinions relationship prediction · Epinions (10-fold CV) | accuracy: 0.9139 -> 0.9474 | Improved |
| add noisy user features via IMCNF to the feature-free low-rank model | MF-ALS · Epinions (10-fold CV) | AUC: 0.9020 -> 0.9506 | Improved |
| add noisy user features via standard IMC (no noise handling) | MF-ALS · Epinions (10-fold CV) | accuracy: 0.9412 -> 0.9139 | Hurt |
| add Eigendigit features to PCP (PCPF instead of PCP) | PCP denoising, linear SVM · MNIST rho_s=0.1 | digit accuracy: 86.33 -> 87.88 | Improved |
| add Eigendigit features to PCP (PCPF instead of PCP) | PCP denoising, kernel SVM · MNIST rho_s=0.1 | digit accuracy: 94.85 -> 95.89 | Improved |
| add Eigendigit features to PCP (PCPF instead of PCP) | PCP denoising, linear SVM · MNIST rho_s=0.3 | digit accuracy: 78.52 -> 79.84 | Improved |
| denoise with PCP instead of using noisy images | MNIST linear SVM · MNIST rho_s=0.1 | digit accuracy: 59.63 -> 86.33 | Improved |
| use noisy label-relevant features with the noise-tolerant model PCPNF-w/Y instead of PCPF-w/Y | PCPF-w/Y · MNIST denoising, rho_f>0.5 | digit accuracy: accuracies much lower than 70 for PCPF-w/Y for rho_f > 0.5 | Improved |
| add perfect label-relevant features (rho_f=0) | PCPF-w/Y and PCPNF-w/Y · MNIST denoising | digit accuracy: accuracy higher than the clean images | Improved |
Closest papers by meaning
- High Rank Matrix Completion With Side Information (AAAI 2018 · similarity 0.90)
- Robust Principal Component Analysis with Side Information (ICML 2016 · similarity 0.89)
- Matrix Completion with Incomplete Side Information via Orthogonal Complement Projection (ICML 2025 · similarity 0.86)
- A Pairwise Pseudo-likelihood Approach for Matrix Completion with Informative Missingness (NeurIPS 2024 · similarity 0.86)
- LERE: Learning-Based Low-Rank Matrix Recovery with Rank Estimation (AAAI 2024 · similarity 0.85)
- RGNMR: A Gauss-Newton method for robust matrix completion with theoretical guarantees (NeurIPS 2025 · similarity 0.84)
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.
