Research map / Learning theory and algorithms / Matrix and tensor methods
Matrix Norm Estimation from a Few Entries
Ashish Khetan, Sewoong Oh
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 |
|---|---|---|---|
| Replace the naive scaled sampled-matrix Schatten norm with the unbiased cycle-counting estimator (de-biasing each pseudograph pattern by its sampling probability) | Schatten k-norm estimation · d = 500, r = 100, k = 5, Erdős-Rényi sampling | absolute relative error | Improved |
| Use the proposed estimator instead of matrix completion (AltMin) to estimate Schatten norms when samples are below the completion threshold | Schatten k-norm estimation · d = 500, r = 500, k = 5 | absolute relative error | Improved |
| Use efficient matrix-multiplication-based counting (O(d^alpha)) of all k-cyclic pseudographs instead of counting only increasing cycles (ablation) | Schatten 7-norm estimation from random vectors (Figure 3 setting) · d = 2048, k = 7 | multiplicative error of Schatten 7-norm estimate | Improved |
| Scaling the sampled matrix's singular values directly (without de-biasing) for Schatten norm estimation when rank is large relative to samples | Scaled sampled matrix baseline · d = 500, r = 100 and r = 500, k = 5 | absolute relative error | Hurt |
Closest papers by meaning
- Spectrum Estimation from a Few Entries (JMLR 2019 · similarity 0.96)
- Schatten Norms in Matrix Streams: Hello Sparsity, Goodbye Dimension (ICML 2020 · similarity 0.92)
- Matrix Norms in Data Streams: Faster, Multi-Pass and Row-Order (ICML 2018 · similarity 0.89)
- Fast exact recovery of noisy matrix from few entries: the infinity norm approach (NeurIPS 2025 · similarity 0.87)
- Spectral Perturbation Bounds for Low-Rank Approximation with Applications to Privacy (NeurIPS 2025 · similarity 0.86)
- Entrywise error bounds for low-rank approximations of kernel matrices (NeurIPS 2024 · 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.
