Research map / Optimization / Nonconvex and smooth optimization
Stochastic Interior-Point Methods for Smooth Conic Optimization with Applications
Chuan He, Zhanwang Deng
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 |
|---|---|---|---|
| add increasing batch size (SIPM-ME+) instead of fixed batch size | SIPM mini-batch estimator · Robust regression, wine-quality (d=10, p=2000) | relative objective value: 0.2860 -> 0.2376 | Improved |
| add increasing batch size (SIPM-ME+) instead of fixed batch size | SIPM mini-batch estimator · Robust regression, wine-quality (d=10, p=2000) | relative stationary error: 1.550e-2 -> 3.611e-3 | Improved |
| add increasing batch size (SIPM-ME+) instead of fixed batch size | SIPM mini-batch estimator · Multi-task relationship learning, ten tasks (m=200) | relative objective value: 0.3985 -> 0.2879 | Improved |
| add increasing batch size (SIPM-ME+) instead of fixed batch size | SIPM mini-batch estimator · Clustering data streams, spam-base (d=100, p=100) | relative stationary error: 3.108e-1 -> 2.950e-2 | Improved |
| use SIPM-RM instead of SALM-RM (interior-point instead of augmented Lagrangian) | recursive momentum method · Robust regression, wine-quality (d=10, p=2000) | relative objective value: 0.2583 -> 0.2306 | Improved |
| use SIPM-RM instead of alternating minimization (AM) | multi-task relationship learning · Multi-task relationship learning, five tasks (m=200) | relative objective value: 0.3535 -> 0.3198 | Improved |
| use SIPM-RM instead of alternating minimization (AM) | multi-task relationship learning · Multi-task relationship learning, ten tasks (m=500) | relative objective value: 0.4253 -> 0.2774 | Improved |
| use stochastic SIPM-RM instead of full-batch IPM-FG | interior-point method · Robust regression, wine-quality (d=10, p=2000) | relative objective value: 0.2359 -> 0.2306 | Improved |
| use stochastic SIPM-RM instead of full-batch IPM-FG | interior-point method · Robust regression, energy-efficiency (d=10, p=5000) | relative objective value: 0.8128 -> 0.8128 | No clear effect |
| use stochastic SIPM-RM instead of full-batch IPM-FG | interior-point method · Clustering data streams, spam-base (d=100, p=100) | relative objective value: -19.15 -> -19.15 | No clear effect |
| use SIPM-RM instead of SIPM-PM (recursive instead of Polyak momentum) | SIPM · Multi-task relationship learning, ten tasks (m=200) | relative objective value: 0.2875 -> 0.2765 | Improved |
| use SIPM-RM instead of SIPM-PM (recursive instead of Polyak momentum) | SIPM · Robust regression, energy-efficiency (d=10, p=5000) | relative stationary error: 8.123e-5 -> 2.167e-6 | Improved |
Closest papers by meaning
- Toward Better Generalization Bounds of Stochastic Optimization for Nonconvex Learning (TPAMI 2025 · similarity 0.87)
- Gradient-Free Methods for Deterministic and Stochastic Nonsmooth Nonconvex Optimization (NeurIPS 2022 · similarity 0.86)
- ProxSARAH: An Efficient Algorithmic Framework for Stochastic Composite Nonconvex Optimization (JMLR 2020 · similarity 0.86)
- PROMISE: Preconditioned Stochastic Optimization Methods by Incorporating Scalable Curvature Estimates (JMLR 2024 · similarity 0.86)
- High-probability complexity bounds for stochastic non-convex minimax optimization (NeurIPS 2024 · similarity 0.86)
- Inexact Augmented Lagrangian Methods for Conic Optimization: Quadratic Growth and Linear Convergence (NeurIPS 2024 · similarity 0.85)
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.
