Research map / Optimization / Nonconvex and smooth optimization
Dual Iterative Hard Thresholding
Xiao-Tong Yuan, Bo Liu, Lezi Wang, Qingshan Liu, Dimitris N. Metaxas
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 DIHT instead of primal IHT and HTP on ill-conditioned linear regression | DIHT · Sparse linear regression, condition number about 50, (d,k)=(500,50) | primal objective and parameter estimation error | Improved |
| increase condition number of the design matrix (ablation) | DIHT · Sparse linear regression, kappa in [1,200] | optimal primal-dual gap | No clear effect |
| increase noise level sigma (ablation) | DIHT · Sparse linear regression, sigma in [10^-3, 50] | optimal primal-dual gap | Hurt |
| increase regularization strength lambda (ablation) | DIHT · Sparse linear regression, lambda0 in {0.01,0.1,1} | optimal primal-dual gap | Improved |
| use DIHT and SDIHT instead of IHT, HTP and SVR-GHT | DIHT/SDIHT · RCV1 and News20, smoothed hinge loss | primal loss versus running time | Improved |
| use stochastic dual updates (SDIHT) instead of batch DIHT | DIHT · RCV1 and News20, smoothed hinge loss | running time and epochs to comparable accuracy and primal-dual gap | Improved |
Closest papers by meaning
- Dual Iterative Hard Thresholding: From Non-convex Sparse Minimization to Non-smooth Concave Maximization (ICML 2017 · similarity 0.96)
- Iterative Hard Thresholding with Adaptive Regularization: Sparser Solutions Without Sacrificing Runtime (ICML 2022 · similarity 0.91)
- Learning Sparse Distributions using Iterative Hard Thresholding (NeurIPS 2019 · similarity 0.91)
- Fast Iterative Hard Thresholding Methods with Pruning Gradient Computations (NeurIPS 2024 · similarity 0.88)
- Optimization over Sparse Support-Preserving Sets: Two-Step Projection with Global Optimality Guarantees (ICML 2025 · similarity 0.87)
- Smoothing Proximal Gradient Methods for Nonsmooth Sparsity Constrained Optimization: Optimality Conditions and Global Convergence (ICML 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.
