# Accelerating Neural Network Training with Distributed Asynchronous and Selective Optimization (DASO)

arXiv: https://arxiv.org/abs/2104.05588 · HTML: https://mlsanity.com/paper/2104.05588/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 23 of its 55 references that are in the map.
- First posted on arXiv 12 Apr 2021.

## Builds on (most cited first, 23 of 23)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Deep Residual Learning for Image Recognition](https://mlsanity.com/paper/1512.03385/) (2015)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [The Cityscapes Dataset for Semantic Urban Scene Understanding](https://mlsanity.com/paper/1604.01685/) (2016)
- [Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour](https://mlsanity.com/paper/1706.02677/) (2017)
- [Generalized Intersection over Union: A Metric and A Loss for Bounding Box Regression](https://mlsanity.com/paper/1902.09630/) (2019)
- [Optimization Methods for Large-Scale Machine Learning](https://mlsanity.com/paper/1606.04838/) (2016)
- [Horovod: fast and easy distributed deep learning in TensorFlow](https://mlsanity.com/paper/1802.05799/) (2018)
- [One weird trick for parallelizing convolutional neural networks](https://mlsanity.com/paper/1404.5997/) (2014)
- [Don't Use Large Mini-Batches, Use Local SGD](https://mlsanity.com/paper/1808.07217/) (2018)
- [Hierarchical Multi-Scale Attention for Semantic Segmentation](https://mlsanity.com/paper/2005.10821/) (2020)
- [Asynchronous Decentralized Parallel Stochastic Gradient Descent](https://mlsanity.com/paper/1710.06952/) (2017)
- [Demystifying Parallel and Distributed Deep Learning: An In-Depth Concurrency Analysis](https://mlsanity.com/paper/1802.09941/) (2018)
- [QSGD: Communication-Efficient SGD via Gradient Quantization and Encoding](https://mlsanity.com/paper/1610.02132/) (2016)
- [Why M Heads are Better than One: Training a Diverse Ensemble of Deep Networks](https://mlsanity.com/paper/1511.06314/) (2015)
- [Slow and Stale Gradients Can Win the Race: Error-Runtime Trade-offs in Distributed SGD](https://mlsanity.com/paper/1803.01113/) (2018)
- [Region Mutual Information Loss for Semantic Segmentation](https://mlsanity.com/paper/1910.12037/) (2019)
- [Staleness-aware Async-SGD for Distributed Deep Learning](https://mlsanity.com/paper/1511.05950/) (2015)
- [Evaluating Modern GPU Interconnect: PCIe, NVLink, NV-SLI, NVSwitch and GPUDirect](https://mlsanity.com/paper/1903.04611/) (2019)
- [Yet Another Accelerated SGD: ResNet-50 Training on ImageNet in 74.7 seconds](https://mlsanity.com/paper/1903.12650/) (2019)
- [Distributed Training of Deep Neural Networks: Theoretical and Practical Limits of Parallel Scalability](https://mlsanity.com/paper/1609.06870/) (2016)
- [HeAT -- a Distributed and GPU-accelerated Tensor Framework for Data Analytics](https://mlsanity.com/paper/2007.13552/) (2020)
- [Accelerating Asynchronous Stochastic Gradient Descent for Neural Machine Translation](https://mlsanity.com/paper/1808.08859/) (2018)

## Built on by (most cited first, 4 of 4)

- [Training Neural Networks from Scratch with Parallel Low-Rank Adapters](https://mlsanity.com/paper/2402.16828/) (2024)
- [TransNAS-TSAD: Harnessing Transformers for Multi-Objective Neural Architecture Search in Time Series Anomaly Detection](https://mlsanity.com/paper/2311.18061/) (2023)
- [AB-Training: A Communication-Efficient Approach for Distributed Low-Rank Learning](https://mlsanity.com/paper/2405.01067/) (2024)
- [PETNet -- Coincident Particle Event Detection using Spiking Neural Networks](https://mlsanity.com/paper/2504.06730/) (2025)

## Abstract (arXiv)

With increasing data and model complexities, the time required to train neural networks has become prohibitively large. To address the exponential rise in training time, users are turning to data parallel neural networks (DPNN) to utilize large-scale distributed resources on computer clusters. Current DPNN approaches implement the network parameter updates by synchronizing and averaging gradients across all processes with blocking communication operations. This synchronization is the central algorithmic bottleneck. To combat this, we introduce the Distributed Asynchronous and Selective Optimization (DASO) method which leverages multi-GPU compute node architectures to accelerate network training. DASO uses a hierarchical and asynchronous communication scheme comprised of node-local and global networks while adjusting the global synchronization rate during the learning process. We show that DASO yields a reduction in training time of up to 34% on classical and state-of-the-art networks, as compared to other existing data parallel training methods.