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

Download:

Current browse context:

cs.LG

Change to browse by:

References & Citations

DBLP - CS Bibliography

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: On the Periodic Behavior of Neural Network Training with Batch Normalization and Weight Decay

Abstract: Training neural networks with batch normalization and weight decay has become a common practice in recent years. In this work, we show that their combined use may result in a surprising periodic behavior of optimization dynamics: the training process regularly exhibits destabilizations that, however, do not lead to complete divergence but cause a new period of training. We rigorously investigate the mechanism underlying the discovered periodic behavior from both empirical and theoretical points of view and analyze the conditions in which it occurs in practice. We also demonstrate that periodic behavior can be regarded as a generalization of two previously opposing perspectives on training with batch normalization and weight decay, namely the equilibrium presumption and the instability presumption.
Comments: Published in NeurIPS 2021. First two authors contributed equally
Subjects: Machine Learning (cs.LG); Machine Learning (stat.ML)
Cite as: arXiv:2106.15739 [cs.LG]
  (or arXiv:2106.15739v3 [cs.LG] for this version)

Submission history

From: Ekaterina Lobacheva Ms [view email]
[v1] Tue, 29 Jun 2021 21:53:14 GMT (5303kb,D)
[v2] Sun, 5 Dec 2021 19:34:20 GMT (25169kb,D)
[v3] Sat, 15 Jan 2022 19:49:44 GMT (27282kb,D)

Link back to: arXiv, form interface, contact.