Research map / Optimization / Nonconvex and smooth optimization
A Composite Randomized Incremental Gradient Method
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Replace SVRG-style estimator with SAGA-style reference-point estimator (C-SAGA) for composite problem | composite finite-sum problem (1) · Theory: nonconvex smooth composition | sample complexity to epsilon-stationarity | No clear effect |
| Use gradient-dominant/strongly convex condition with C-SAGA | composite finite-sum problem · Linear convergence regime | sample complexity O((n+kappa n^{2/3}) log(1/eps)) | Helped |
| C-SAGA vs VRSC-PG on risk-averse portfolio | C-SAGA · Risk-averse portfolio optimization, d=500, n=5000 | sample efficiency (gradient norm / objective gap vs samples) | Helped |
| Use C-SAGA instead of ASC-PG/VRSC-PG/SCGD/ASCGD for MDP policy evaluation | C-SAGA · MDP policy evaluation, S=10 and S=100 | objective gap and gradient norm vs samples | Helped |
| Single sample set (no double sampling) vs Reddi et al. sharper bound (ablation) | C-SAGA reference-point analysis · Theory | Lemma 4 bound constant | No clear effect |
Closest papers by meaning
- A Stochastic Composite Gradient Method with Incremental Variance Reduction (NeurIPS 2019 · similarity 0.92)
- On Convergence of Incremental Gradient for Non-convex Smooth Functions (ICML 2024 · similarity 0.86)
- Decentralized Gradient-Free Methods for Stochastic Non-smooth Non-convex Optimization (AAAI 2024 · similarity 0.86)
- A Bregman Proximal Stochastic Gradient Method with Extrapolation for Nonconvex Nonsmooth Problems (AAAI 2024 · similarity 0.86)
- Efficient Continual Finite-Sum Minimization (ICLR 2024 · similarity 0.86)
- Incremental Gradient Descent with Small Epoch Counts is Surprisingly Slow on Ill-Conditioned Problems (ICML 2025 · 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.