The DOI matches its Crossref record; 26 of the 26 references it cites were found in a registry or an index.
Parallel Restarted SGD with Faster Convergence and Less Communication: Demystifying Why Model Averaging Works for Deep Learning
Open accessAbstract
In distributed training of deep neural networks, parallel minibatch SGD is widely used to speed up the training process by using multiple workers. It uses multiple workers to sample local stochastic gradients in parallel, aggregates all gradients in a single server to obtain the average, and updates each worker’s local model using a SGD update with the averaged gradient. Ideally, parallel mini-batch SGD can achieve a linear speed-up of the training time (with respect to the number of workers) compared with SGD over a single worker. However, such linear scalability in practice is significantly limited by the growing demand for gradient communication as more workers are involved. Model averaging, which periodically averages individual models trained over parallel workers, is another common practice used for distributed training of deep neural networks since (Zinkevich et al. 2010) (McDonald, Hall, and Mann 2010). Compared with parallel mini-batch SGD, the communication overhead of model averaging is significantly reduced. Impressively, tremendous experimental works have verified that model averaging can still achieve a good speed-up of the training time as long as the averaging interval is carefully controlled. However, it remains a mystery in theory why such a simple heuristic works so well. This paper provides a thorough and rigorous theoretical study on why model averaging can work as well as parallel mini-batch SGD with significantly less communication overhead.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Large Scale Distributed Deep Networks
Dean, J. B., Corrado, G. S., Monga, R., Chen, K., Devin, M., Mao, M., Ranzato, M., Senior, A., Tucker, P. A., Yang, K., et al. (2012) · neural information processing systems · Cited by 4,489
Accurate, Large Minibatch SGD: Training ImageNet in 1 Hour
Goyal, P., Dollár, P., Girshick, R., Noordhuis, P., Wesolowski, L., Kyrola, A., Tulloch, A., Jia, Y., & He, K. (2017) · arXiv · doi:10.48550/arxiv.1706.02677 · Cited by 2,294
Parallelized Stochastic Gradient Descent
Zinkevich, M., Weimer, M., Li, L., & Smola, A. J. (2010) · Cited by 1,124
Stochastic First- and Zeroth-Order Methods for Nonconvex Stochastic Programming
Ghadimi, S., & Lan, G. (2013) · SIAM Journal on Optimization · doi:10.1137/120880811 · Cited by 1,034
QSGD: Communication-Efficient SGD via Gradient Quantization and Encoding
Alistarh, D., Grubic, D., Li, J., Tomioka, R., & Vojnović, M. (2016) · London School of Economics and Political Science Research Online (London School of Economics and Political Science) · doi:10.48550/arxiv.1610.02132 · Cited by 859
Communication Efficient Distributed Machine Learning with the Parameter Server
Li, M., Andersen, D. G., Smola, A. J., & Yu, K. (2014) · neural information processing systems · Cited by 715
TernGrad: Ternary Gradients to Reduce Communication in Distributed Deep Learning
Wen, W., Xu, C., Yan, F., Wu, C., Wang, Y., Chen, Y., & Li, H. H. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1705.07878 · Cited by 622
Scalable distributed DNN training using commodity GPU cloud computing
Strom, N. (2015) · Interspeech 2015 · doi:10.21437/interspeech.2015-354 · Cited by 521
Optimal Distributed Online Prediction using Mini-Batches
Dekel, O., Gilad-Bachrach, R., Shamir, O., & Xiao, L. (2010) · arXiv (Cornell University) · doi:10.48550/arxiv.1012.1367 · Cited by 512
Asynchronous Decentralized Parallel Stochastic Gradient DescentPreprint
Lian, X., Zhang, W., Zhang, C., & Liu, J. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1710.06952 · Cited by 444
Seide, F., Fu, H., Droppo, J., Li, G., & Yu, D. (2014) · Interspeech 2014 · doi:10.21437/interspeech.2014-274 · Cited by 443
Lian, X., Zhang, C., Zhang, H., Hsieh, C., Zhang, W., & Liu, J. (2017) · arXiv (Cornell University) · doi:10.48550/arxiv.1705.09056 · Cited by 395
Distributed Training Strategies for the Structured Perceptron
McDonald, R., Hall, K., & Mann, G. (2010) · North American Chapter of the Association for Computational Linguistics · Cited by 335
Communication-efficient algorithms for statistical optimization
Zhang, Y., Duchi, J. C., & Wainwright, M. J. (2012) · 2012 IEEE 51st IEEE Conference on Decision and Control (CDC) · doi:10.1109/cdc.2012.6426691 · Cited by 283
Perturbed Iterate Analysis for Asynchronous Stochastic Optimization
Mania, H., Pan, X., Papailiopoulos, D., Recht, B., Ramchandran, K., & Jordan, M. I. (2017) · SIAM Journal on Optimization · doi:10.1137/16m1057000 · Cited by 246
Collaborative Deep Learning in Fixed Topology NetworksPreprint
Jiang, Z., Balu, A., Hegde, C., & Sarkar, S. (2017) · Neural Information Processing Systems · doi:10.48550/arxiv.1706.07880 · Cited by 206
Communication-efficient algorithms for statistical optimization
(2013) · Journal of Machine Learning Research · doi:10.5555/2567709.2567769 · Cited by 195
Don't Use Large Mini-Batches, Use Local SGD
Lin, T., Stich, S. U., Patel, K. K., & Jaggi, M. (2018) · arXiv (Cornell University) · doi:10.48550/arxiv.1808.07217 · Cited by 193
Chen, K., & Huo, Q. (2016) · IEEE International Conference on Acoustics Speech and Signal Processing · doi:10.1109/icassp.2016.7472805 · Cited by 154
Parallel training of DNNs with Natural Gradient and Parameter Averaging
Povey, D., Zhang, X., & Khudanpur, S. (2014) · arXiv · doi:10.48550/arxiv.1410.7455 · Cited by 148
Efficient Decentralized Deep Learning by Dynamic Model Averaging
Kamp, M., Adilova, L., Sicking, J., Hüger, F., Schlicht, P., Wirtz, T., & Wrobel, S. (2019) · Lecture notes in computer science · doi:10.1007/978-3-030-10925-7_24 · Cited by 146
Communication Quantization for Data-Parallel Training of Deep Neural Networks
Dryden, N., Moon, T., Jacobs, S. A., & Essen, B. V. (2016) · doi:10.1109/mlhpc.2016.004 · Cited by 144
Parallel SGD: When does averaging help?Preprint
Zhang, J., De, C., Mitliagkas, I., & Ré, C. (2016) · arXiv (Cornell University) · doi:10.48550/arxiv.1606.07365 · Cited by 111
Communication quantization for data-parallel training of deep neural networks
Dryden, N., Jacobs, S. A., Moon, T., & Essen, B. V. (2016) · IEEE International Conference on High Performance Computing, Data, and Analytics · doi:10.5555/3018874.3018875 · Cited by 111
Experiments on Parallel Training of Deep Neural Network using Model AveragingPreprint
Su, H., & Chen, H. (2015) · arXiv (Cornell University) · doi:10.48550/arxiv.1507.01239 · Cited by 76
(2018) · arXiv (Cornell University) · Cited by 42
Cited by
Federated learning: Challenges, methods, and future directions
Li, T., Sahu, A. K., Talwalkar, A., & Smith, V. (2020) · IEEE Signal Processing Magazine · doi:10.1109/msp.2020.2975749 · Cited by 6,210
Adaptive Federated Learning in Resource Constrained Edge Computing Systems
Wang, S., Tuor, T., Salonidis, T., Leung, K. K., Makaya, C., He, T., & Chan, K. (2019) · IEEE Journal on Selected Areas in Communications · doi:10.1109/jsac.2019.2904348 · Cited by 2,232
A survey on federated learning
Zhang, C., Xie, Y., Bai, H., Yu, B., Li, W., & Gao, Y. (2021) · Knowledge-Based Systems · doi:10.1016/j.knosys.2021.106775 · Cited by 1,899
SCAFFOLD: Stochastic Controlled Averaging for Federated Learning
Karimireddy, S. P., Kale, S., Mohri, M., Reddi, S. J., Stich, S. U., & Suresh, A. T. (2019) · arXiv · doi:10.48550/arxiv.1910.06378 · Cited by 1,734
“Several analyses bound this drift by using a very small step-size and assuming that the local updates admit bounded magnitude (Wang et al., 2019; Yu et al., 2019; Li et al., 2019b).”
BackgroundOn the Convergence of FedAvg on Non-IID Data
Li, X., Huang, K., Yang, W., Wang, S., & Zhang, Z. (2019) · arXiv · doi:10.48550/arxiv.1907.02189 · Cited by 1,575
“Assumptions 3 and 4 have been made by the works (Zhang et al., 2013; Stich, 2018; Stich et al., 2018; Yu et al., 2019).”
Method
Metadata
- Authors
- Yu, H., Yang, S., & Zhu, S.
- Venue
- Proceedings of the AAAI Conference on Artificial Intelligence
- Published
- 2019-07-17
- Volume
- 33
- Issue
- 01
- Pages
- 5693-5700
- ISSN
- 2159-5399 (ISSN-L)
- 2374-3468
- OpenAlex
- W2900182564
- Cited by
- 497
- References
- 26
- FWCI
- 26.54
- Type
- conference paper
- Language
- en
- Open access
- diamond
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.