Research map / Learning theory and algorithms / Matrix and tensor methods
Non-Convex Projected Gradient Descent for Generalized Low-Rank Tensor Regression
Han Chen, Garvesh Raskutti, Ming Yuan
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 non-convex PGD instead of convex regularization | Low-rank tensor regression · Case 6a Gaussian low CP rank, high SNR | normalized rmse: 0.28 -> 0.11 | Improved |
| Use non-convex PGD instead of convex regularization | Low-rank tensor regression · Case 6c Poisson low CP rank, high SNR | normalized rmse: 0.57 -> 0.09 | Improved |
| Use non-convex PGD instead of convex regularization | Low-rank tensor regression · Case 8c Poisson sparse low-rank slices, high SNR | normalized rmse: 0.79 -> 0.13 | Improved |
| Use non-convex PGD instead of convex regularization | Low-rank tensor regression · Case 8a Gaussian sparse low-rank slices, high SNR | normalized rmse: 0.12 -> 0.08 | Improved |
| Use non-convex PGD instead of convex regularization | Low-rank tensor regression · Case 7a Gaussian low Tucker rank, low SNR | normalized rmse: 0.51 -> 0.28 | Improved |
| Use a larger assumed rank r' or sparsity s' in the PGD projection (misspecification) (ablation) | Non-convex PGD · Cases 1a-3a Gaussian third-order tensors, p=50^3 | converged rmse | Hurt |
| Use a larger step size eta (below the divergence tolerance) (ablation) | Non-convex PGD · Cases 1a-3a Gaussian third-order tensors, p=50^3 | convergence speed | Improved |
| Use too large a step size eta (ablation) | Non-convex PGD · Cases 1a-3a Gaussian third-order tensors, p=50^3 | convergence | Hurt |
| Use the relaxed (misspecified) projection instead of the true rank (ablation) | Non-convex PGD · Cases 1b-1c logistic and Poisson low CP rank | convergence rmse | Hurt |
Closest papers by meaning
- Boosted Sparse and Low-Rank Tensor Regression (NeurIPS 2018 · similarity 0.91)
- High-Dimensional Tensor Regression With Oracle Properties (ICML 2025 · similarity 0.90)
- Near Optimal Sketching of Low-Rank Tensor Regression (NeurIPS 2017 · similarity 0.90)
- Low-Rank Tensor Learning by Generalized Nonconvex Regularization (TPAMI 2026 · similarity 0.90)
- Low-Rank Tensor Transitions (LoRT) for Transferable Tensor Regression (ICML 2025 · similarity 0.88)
- Non-Convex Tensor Recovery from Local Measurements (AAAI 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.
