Research map / Optimization / Nonconvex and smooth optimization
Decentralized Nonconvex Optimization under Heavy-Tailed Noise: Normalization and Optimal Convergence
Shuhua Yu, Dusan Jakovetic, Soummya Kar
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 GT-NSGDm instead of DSGD-Clip | decentralized training · Multi30k GPT, ring | validation loss: 5.633 -> 0.258 | Improved |
| use GT-NSGDm instead of SClip-EF-Network | decentralized training · Multi30k GPT, ring | validation loss: 5.653 -> 0.258 | Improved |
| use GT-NSGDm instead of DSGD-GClip (no theory) | decentralized training · Multi30k GPT, ring | validation loss: 0.253 -> 0.258 | No clear effect |
| use GT-NSGDm instead of GT-Adam | decentralized training · Multi30k GPT, complete graph | validation loss: 0.524 -> 0.282 | Improved |
| weaker connectivity (ring vs complete graph) (ablation) | GT-NSGDm · synthetic regression, Levy noise | final error | No clear effect |
| add gradient tracking and normalization instead of plain DSGD | decentralized SGD · synthetic regression, heavy-tailed noise | estimation error | Improved |
Closest papers by meaning
- Stability and Generalization of Nonconvex Optimization with Heavy-Tailed Noise (ICML 2026 · similarity 0.92)
- Nonconvex Stochastic Optimization under Heavy-Tailed Noises: Optimal Convergence without Gradient Clipping (ICLR 2025 · similarity 0.91)
- Nonconvex Decentralized Stochastic Bilevel Optimization under Heavy-Tailed Noise (NeurIPS 2026 · similarity 0.91)
- Efficient Distributed Optimization under Heavy-Tailed Noise (ICML 2025 · similarity 0.90)
- Stochastic Zeroth-Order Optimization Under Heavy-Tailed Noise (NeurIPS 2026 · similarity 0.89)
- Error Analysis Affected by Heavy-Tailed Gradients for Non-Convex Pairwise Stochastic Gradient Descent (AAAI 2025 · similarity 0.89)
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.
