Research map / Optimization / Nonconvex and smooth optimization
Adaptive Proximal Average Based Variance Reducing Stochastic Methods for Optimization with Composite Regularization
Jingchang Liu, Linli Xu, Junliang Guo, Xin Sheng
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 adaptive decreasing stepsize with large initial value (APA) instead of small fixed stepsize (PA) | PA-SVRG · overlapping group lasso, K in {5,10,20,50} | objective value vs iterations | Improved |
| use adaptive decreasing stepsize with large initial value (APA) instead of small fixed stepsize (PA) | PA-SAGA · graph-guided fused lasso, four LIBSVM datasets | objective value vs iterations | Improved |
| replace SVRG-ADMM with APA-SVRG / APA-SAGA | APA methods · overlapping group lasso | convergence speed and implementation simplicity | Improved |
| compare APA-SVRG/APA-SAGA with SVRG-ADMM | APA methods · graph-guided fused lasso | approach to optimum | No clear effect |
| add variance reduction to PA-based stochastic method (PA-ASGD without VR) | PA-ASGD · graph-guided fused lasso | objective value | Improved |
| use APA methods instead of PA-ASGD | APA methods · overlapping group lasso | objective value | Improved |
| use adaptive stepsize so iterates converge to optimum | APA-SVRG / APA-SAGA · graph-guided fused lasso | convergence to optimal point | Improved |
| raise desired accuracy epsilon for PA methods (1e-4 to 1e-5) | PA-SVRG / PA-SAGA · overlapping group lasso K=50 | required stepsize | Hurt |
Closest papers by meaning
- Adaptive Proximal Average Approximation for Composite Convex Minimization (AAAI 2017 · similarity 0.88)
- Convergence Rate of the Last Iterate of Stochastic Proximal Algorithms (ICML 2026 · similarity 0.88)
- Asynchronous Stochastic Proximal Optimization Algorithms with Variance Reduction (AAAI 2017 · similarity 0.88)
- RandProx: Primal-Dual Optimization Algorithms with Randomized Proximal Updates (ICLR 2023 · similarity 0.87)
- Double Variance Reduction: A Smoothing Trick for Composite Optimization Problems without First-Order Gradient (ICML 2024 · similarity 0.86)
- Non-convex Stochastic Composite Optimization with Polyak Momentum (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.
