We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:

Download:

Current browse context:

stat.ML

Change to browse by:

References & Citations

Bookmark

(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: Implicit Gradient Alignment in Distributed and Federated Learning

Abstract: A major obstacle to achieving global convergence in distributed and federated learning is the misalignment of gradients across clients, or mini-batches due to heterogeneity and stochasticity of the distributed data. In this work, we show that data heterogeneity can in fact be exploited to improve generalization performance through implicit regularization. One way to alleviate the effects of heterogeneity is to encourage the alignment of gradients across different clients throughout training. Our analysis reveals that this goal can be accomplished by utilizing the right optimization method that replicates the implicit regularization effect of SGD, leading to gradient alignment as well as improvements in test accuracies. Since the existence of this regularization in SGD completely relies on the sequential use of different mini-batches during training, it is inherently absent when training with large mini-batches. To obtain the generalization benefits of this regularization while increasing parallelism, we propose a novel GradAlign algorithm that induces the same implicit regularization while allowing the use of arbitrarily large batches in each update. We experimentally validate the benefits of our algorithm in different distributed and federated learning settings.
Comments: AAAI Conference on Artificial Intelligence, 2022
Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:2106.13897 [cs.LG]
  (or arXiv:2106.13897v3 [cs.LG] for this version)

Submission history

From: Yatin Dandi [view email]
[v1] Fri, 25 Jun 2021 22:01:35 GMT (627kb,D)
[v2] Fri, 29 Oct 2021 19:33:02 GMT (1097kb,D)
[v3] Mon, 13 Dec 2021 00:18:10 GMT (2208kb,D)

Link back to: arXiv, form interface, contact.