Research map / Optimization / Optimizers
Stochastic Iterative Hard Thresholding for Graph-structured Sparsity Optimization
Baojian Zhou, Feng Chen, Yiming Ying
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 GraphStoIHT instead of GraphIHT | Graph-structured logistic regression · Breast cancer dataset | AUC: 0.714±0.06 -> 0.715±0.06 | Improved |
| Use graph-structured stochastic hard thresholding instead of StoIHT | Graph-structured logistic regression · Breast cancer dataset | AUC: 0.708±0.06 -> 0.715±0.06 | Improved |
| Use GraphStoIHT instead of GraphIHT | Breast cancer gene identification · Breast cancer-related genes | related genes found: 20% -> 24% | Improved |
| Use GraphStoIHT instead of StoIHT | Breast cancer gene identification · Breast cancer-related genes | related genes found: 16% -> 24% | Improved |
| Use GraphStoIHT instead of IHT | Breast cancer gene identification · Breast cancer-related genes | related genes found: 8% -> 24% | Improved |
| Increase block size to at least the sparsity level (ablation) | GraphStoIHT · Synthetic graph sparse linear regression | recovery within 30 epochs | Improved |
| Use a suitable intermediate block size (reported as between 32 and 64) (ablation) | GraphStoIHT · Synthetic graph sparse linear regression | convergence speed | Improved |
| Use a relatively large learning rate (η ≥ 1.5) (ablation) | GraphStoIHT · Synthetic graph sparse linear regression | convergence | Improved |
| Use GraphStoIHT rather than IHT or GraphIHT | Graph sparse linear regression · Synthetic random-walk graphs | observations needed for successful recovery | Improved |
| Use GraphStoIHT rather than StoIHT | Graph sparse linear regression · Synthetic random-walk graphs | observations needed for successful recovery | Improved |
| Use GraphStoIHT with block sizes between 4 and 8 (ablation) | Graph sparse linear regression · Synthetic data under noiseless and noisy conditions | observations required for recovery | Improved |
Closest papers by meaning
- An Effective Hard Thresholding Method Based on Stochastic Variance Reduction for Nonconvex Sparse Learning (AAAI 2020 · similarity 0.90)
- Dual Iterative Hard Thresholding: From Non-convex Sparse Minimization to Non-smooth Concave Maximization (ICML 2017 · similarity 0.89)
- Iterative Hard Thresholding with Adaptive Regularization: Sparser Solutions Without Sacrificing Runtime (ICML 2022 · similarity 0.89)
- Sample Efficient Stochastic Gradient Iterative Hard Thresholding Method for Stochastic Sparse Linear Regression with Limited Attribute Observation (NeurIPS 2018 · similarity 0.88)
- Sparse Convex Optimization via Adaptively Regularized Hard Thresholding (ICML 2020 · similarity 0.88)
- Smoothing Proximal Gradient Methods for Nonsmooth Sparsity Constrained Optimization: Optimality Conditions and Global Convergence (ICML 2024 · 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.
