Research map / Optimization / Optimizers
Asymptotic Behaviors of Projected Stochastic Approximation: A Jump Diffusion Perspective
Jiadong Liang, Yuze Han, Xiang Li, Zhihua Zhang
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 |
|---|---|---|---|
| Vary beta within [0, 0.5) (less frequent projection) | LPSA · Synthetic IID, MSE convergence rate | MSE slope | No clear effect |
| Use larger beta and smaller alpha when beta in (1/2, 1) | LPSA · Synthetic IID, MSE curves | smoothness of MSE lines | Improved |
| Use frequent projection with alpha=1 and beta in {0, 0.2} | LPSA · Synthetic IID, rescaled last iterates | limiting distribution of rescaled iterate | Improved |
| Use occasional projection with alpha=0.8 and beta=0.6 | LPSA · Synthetic (1, 1), rescaled iterate trajectories | bias of rescaled iterate | Hurt |
| Set beta=1 (p_n proportional to eta_n) for a specific A | LPSA · Quadratic objective, linear constraint | E||u_n - x*||^2 convergence | Hurt |
| Add probabilistic projection with beta in [0, 1/2) in place of projecting every step | Projected SGD · Theory, asymptotic variance | asymptotic variance | No clear effect |
Closest papers by meaning
- Scaffold with Stochastic Gradients: New Analysis with Linear Speed-Up (ICML 2025 · similarity 0.86)
- Stabilized Proximal-Point Methods for Federated Optimization (NeurIPS 2024 · similarity 0.86)
- Methods with Local Steps and Random Reshuffling for Generally Smooth Non-Convex Federated Optimization (ICLR 2025 · similarity 0.85)
- Exact and Linear Convergence for Federated Learning under Arbitrary Client Participation is Attainable (NeurIPS 2025 · similarity 0.85)
- SCAFFLSA: Taming Heterogeneity in Federated Linear Stochastic Approximation and TD Learning (NeurIPS 2024 · similarity 0.85)
- Accelerating Distributed Stochastic Optimization via Self-Repellent Random Walks (ICLR 2024 · similarity 0.84)
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.
