Papers › AutoDropout: Learning Dropout Patterns to Regularize Deep Networks

AutoDropout: Learning Dropout Patterns to Regularize Deep Networks

5 Jan 2021arXiv:2101.01761archive 2025-07-28

Hieu Pham, Quoc V. Le

Neural networks are often over-parameterized and hence benefit from aggressive regularization. Conventional regularization methods, such as Dropout or weight decay, do not leverage the structures of the network's inputs and hidden states. As a result, these conventional methods are less effective than methods that leverage the structures, such as SpatialDropout and DropBlock, which randomly drop the values at certain contiguous areas in the hidden states and setting them to zero. Although the locations of dropout areas random, the patterns of SpatialDropout and DropBlock are manually designed and fixed. Here we propose to learn the dropout patterns. In our method, a controller learns to generate a dropout pattern at every channel and layer of a target network, such as a ConvNet or a Transformer. The target network is then trained with the dropout pattern, and its resulting validation performance is used as a signal for the controller to learn from. We show that this method works well for both image recognition on CIFAR-10 and ImageNet, as well as language modeling on Penn Treebank and WikiText-2. The learned dropout patterns also transfers to different tasks and datasets, such as from language model on Penn Treebank to Engligh-French translation on WMT 2014. Our code will be available.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

google-research/google-research officialmentioned in papertf report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Image ClassificationLanguage ModelingLanguage ModellingMachine Translation

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification CIFAR-10 WRN-28-10+AutoDropout+RandAugment Percentage correct 97.9 #64 of 265 Archive leaderboard report
Image Classification CIFAR-10 AutoDropout Percentage correct 96.8 #99 of 265 Archive leaderboard report
Image Classification ImageNet ResNet-50+AutoDropout+RandAugment Top 1 Accuracy 80.3% #708 of 1060 Archive leaderboard report
Image Classification ImageNet ResNet-50 Top 1 Accuracy 78.7% #810 of 1060 Archive leaderboard report
Image Classification ImageNet EfficientNet-B0 Top 1 Accuracy 77.5% #872 of 1060 Archive leaderboard report
Image Classification ImageNet-10 ResNet-50 + UDA+AutoDropout Top 1 Accuracy 72.9 #1 of 2 Archive leaderboard report
Image Classification cifar-10,4000 WRN-28-2 + UDA+AutoDropout Percentage error 4.2 #1 of 1 Archive leaderboard report
Language Modelling Penn Treebank (Word Level) Transformer-XL + AutoDropout Test perplexity 54.9 #23 of 43 Archive leaderboard report
Language Modelling Penn Treebank (Word Level) Transformer-XL + AutoDropout Validation perplexity 58.1 #23 of 43 Archive leaderboard report
Machine Translation IWSLT2014 German-English TransformerBase + AutoDropout BLEU score 35.8 #19 of 34 Archive leaderboard report
Machine Translation WMT2014 English-French TransformerBase + AutoDropout BLEU score 40 #35 of 57 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Absolute Position EncodingsAdamAttentionAutoDropoutBPEDense ConnectionsDropBlockDropoutLabel SmoothingLayer NormalizationLinear LayerMulti-Head AttentionPosition-Wise Feed-Forward LayerResidual ConnectionSoftmaxSpatialDropoutTransformer

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections