Research map / Optimization / Distributed and compressed
Asynchronous Decentralized Parallel Stochastic Gradient Descent
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 |
|---|---|---|---|
| Use asynchronous decentralized updates instead of AllReduce-SGD | VGG · CIFAR10, 16 workers | testing accuracy: 87.04% -> 88.58% | Helped |
| Use asynchronous decentralized updates instead of AllReduce-SGD | ResNet-20 · CIFAR10, 16 workers | testing accuracy: 90.72% -> 91.49% | Helped |
| Use asynchronous decentralized updates instead of D-PSGD | ResNet-50 · ImageNet-1K, 128 workers, 90 epochs | testing accuracy: 70.90% -> 74.23% | Helped |
| Use asynchronous decentralized updates instead of AllReduce-SGD | ResNet-50 · ImageNet-1K, 128 workers, 90 epochs | testing accuracy: 74.78% -> 74.23% | No clear effect |
| Use asynchronous decentralized updates rather than a synchronous method | ResNet-20 · CIFAR10, one worker slowed 100X | time/epoch (sec): 100.4 -> 1.33 | Helped |
| Use asynchronous decentralized updates rather than a synchronous method | ResNet-20 · CIFAR10, one worker slowed 100X | speedup: 0.18 -> 13.56 | Helped |
| Use asynchronous decentralized updates rather than synchronous AllReduce-SGD | ResNet-50 · ImageNet, 64 workers on shared network | time/epoch (sec): over 1000 -> 264 | Helped |
| Use AD-PSGD rather than EAMSGD | VGG and ResNet-20 · CIFAR10, 16 workers | training loss versus epochs | Helped |
Closest papers by meaning
- Sharper Convergence Guarantees for Asynchronous SGD for Distributed and Federated Learning (NeurIPS 2022 · similarity 0.93)
- Fast Asynchronous Parallel Stochastic Gradient Descent: A Lock-Free Approach with Convergence Guarantee (AAAI 2016 · similarity 0.92)
- Asynchronous Distributed Semi-Stochastic Gradient Optimization (AAAI 2016 · similarity 0.92)
- Ringmaster ASGD: The First Asynchronous SGD with Optimal Time Complexity (ICML 2025 · similarity 0.91)
- Rescaled Asynchronous SGD: Optimal Distributed Optimization under Data and System Heterogeneity (NeurIPS 2026 · similarity 0.91)
- Stability and Generalization of Asynchronous SGD: Sharper Bounds Beyond Lipschitz and Smoothness (NeurIPS 2024 · similarity 0.90)
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.