Research map / Learning theory and algorithms / Matrix and tensor methods
A Tight Bound of Hard Thresholding
Jie Shen, Ping Li
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 HT-SVRG (stochastic) instead of batch PGD | hard thresholding sparse recovery · Sparse recovery phase transition | percentage of successful recovery | No clear effect |
| use PGD with flexible sparsity k and step size instead of IHT (k=K, eta=1) | batch hard thresholding · Sparse recovery phase transition | percentage of successful recovery | Improved |
| raise sparsity parameter k above the true sparsity K (ablation) | HT-SVRG · Sparse recovery, K=4, n=100 | signal recovery | Improved |
| increase update frequency m (larger inner loop) (ablation) | HT-SVRG · Sparse recovery, K=4, n=100 | convergence speed per full gradient | Mixed |
| increase step size eta to 3 or above (ablation) | HT-SVRG · Sparse recovery, K=4, n=100 | convergence: divergence when eta >= 3 | Hurt |
| increase m for very small step size eta (ablation) | HT-SVRG · Sparse recovery, small step | convergence | Improved |
Closest papers by meaning
- Sparse Convex Optimization via Adaptively Regularized Hard Thresholding (ICML 2020 · similarity 0.92)
- Global and Quadratic Convergence of Newton Hard-Thresholding Pursuit (JMLR 2021 · similarity 0.91)
- Partial Hard Thresholding: Towards A Principled Analysis of Support Recovery (NeurIPS 2017 · similarity 0.90)
- Iterative Hard Thresholding with Adaptive Regularization: Sparser Solutions Without Sacrificing Runtime (ICML 2022 · similarity 0.89)
- The Price of Sparsity: Sufficient Conditions for Sparse Recovery using Sparse and Sparsified Measurements (NeurIPS 2025 · similarity 0.89)
- Exact Recovery of Sparse Binary Vectors from Generalized Linear Measurements (ICML 2025 · similarity 0.89)
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.
