The DOI matches its Crossref record; 40 of the 40 references it cites were found in a registry or an index.
Understanding the Disharmony Between Dropout and Batch Normalization by Variance Shift
Open accessAbstract
This paper first answers the question ``why do the two most powerful techniques Dropout and Batch Normalization (BN) often lead to a worse performance when they are combined together in many modern neural networks, but cooperate well sometimes as in Wide ResNet (WRN)?'' in both theoretical and empirical aspects. Theoretically, we find that Dropout shifts the variance of a specific neural unit when we transfer the state of that network from training to test. However, BN maintains its statistical variance, which is accumulated from the entire learning procedure, in the test phase. The inconsistency of variances in Dropout and BN (we name this scheme ``variance shift'') causes the unstable numerical behavior in inference that leads to erroneous predictions finally. Meanwhile, the large feature dimension in WRN further reduces the ``variance shift'' to bring benefits to the overall performance. Thorough experiments on representative modern convolutional networks like DenseNet, ResNet, ResNeXt and Wide ResNet confirm our findings. According to the uncovered mechanism, we get better understandings in the combination of these two techniques and summarize guidelines for better practices.
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
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474
Densely Connected Convolutional Networks
Huang, G., Liu, Z., Van Der Maaten, L., & Weinberger, K. Q. (2017) · 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2017.243 · Cited by 48,566
Squeeze-and-Excitation Networks
Hu, J., Shen, L., & Sun, G. (2018) · 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2018.00745 · Cited by 35,667
(2014) · Cited by 33,419
Rethinking the Inception Architecture for Computer Vision
Szegedy, C., Vanhoucke, V., Ioffe, S., Shlens, J., & Wojna, Z. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.308 · Cited by 32,138
Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
Ioffe, S., & Szegedy, C. (2015) · arXiv · doi:10.48550/arxiv.1502.03167 · Cited by 27,818
The Elements of Statistical Learning
Hastie, T., Friedman, J., & Tibshirani, R. (2001) · Springer Series in Statistics · doi:10.1007/978-0-387-21606-5 · Cited by 24,162
Learning Multiple Layers of Features from Tiny Images
Krizhevsky, A. (2024) · doi:10.57702/zp44cu3g · Cited by 23,395
MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applications
Howard, A. G., Zhu, M., Chen, B., Kalenichenko, D., Wang, W., Weyand, T., Andreetto, M., & Adam, H. (2017) · arXiv · doi:10.48550/arxiv.1704.04861 · Cited by 19,015
(2010) · Cited by 14,134
Cited by
ALBERT: A Lite BERT for Self-supervised Learning of Language Representations
Lan, Z., Chen, M., Goodman, S., Gimpel, K., Sharma, P., & Soricut, R. (2019) · arXiv · doi:10.48550/arxiv.1909.11942 · Cited by 4,308
“, 2017) and theoretical (Li et al., 2019) evidence showing that a combination of batch normalization and dropout in Convolutional Neural Networks may have harmful results.”
BackgroundDeep learning for cellular image analysis
Moen, E., Bannon, D., Kudo, T., Graf, W., Covert, M., & Van Valen, D. (2019) · Nature Methods · doi:10.1038/s41592-019-0403-1 · Cited by 1,225
Dropout vs. batch normalization: an empirical study of their impact to deep learning
Garbin, C., Zhu, X., & Marques, O. (2020) · Multimedia Tools and Applications · doi:10.1007/s11042-019-08453-9 · Cited by 571
Normalization Techniques in Training DNNs: Methodology, Analysis and Application
Huang, L., Qin, J., Zhou, Y., Zhu, F., Liu, L., & Shao, L. (2023) · IEEE Transactions on Pattern Analysis and Machine Intelligence · doi:10.1109/tpami.2023.3250241 · Cited by 537
Larger Norm More Transferable: An Adaptive Feature Norm Approach for Unsupervised Domain Adaptation
Xu, R., Li, G., Yang, J., & Lin, L. (2019) · 2019 IEEE/CVF International Conference on Computer Vision (ICCV) · doi:10.1109/iccv.2019.00151 · Cited by 508
Self-citationSalsaNext: Fast, Uncertainty-Aware Semantic Segmentation of LiDAR Point Clouds
Cortinhal, T., Tzelepis, G., & Erdal Aksoy, E. (2020) · Lecture Notes in Computer Science · doi:10.1007/978-3-030-64559-5_16 · Cited by 430
A cloud detection algorithm for satellite imagery based on deep learning
Jeppesen, J. H., Jacobsen, R. H., Inceoglu, F., & Toftegaard, T. S. (2019) · Remote Sensing of Environment · doi:10.1016/j.rse.2019.03.039 · Cited by 402
A comprehensive survey on regularization strategies in machine learning
Tian, Y., & Zhang, Y. (2022) · Information Fusion · doi:10.1016/j.inffus.2021.11.005 · Cited by 298
RLDS: An explainable residual learning diagnosis system for fetal congenital heart disease
Qiao, S., Pang, S., Luo, G., Pan, S., Yu, Z., Chen, T., & Lv, Z. (2022) · Future Generation Computer Systems · doi:10.1016/j.future.2021.10.001 · Cited by 42
DiffLoad: Uncertainty Quantification in Electrical Load Forecasting With the Diffusion Model
Wang, Z., Wen, Q., Zhang, C., Sun, L., & Wang, Y. (2025) · IEEE Transactions on Power Systems · doi:10.1109/tpwrs.2024.3449032 · Cited by 34
“However,ithasbeenproventhatits forecastingperformancewasunstableduetoinconsistenciesin thetrainingandtestingprocesses[28].”
Decoding visual motions from EEG using attention based RNN.
Yang, D., Liu, Y., Zhou, Z., Yu, Y., & Liang, X. (2020) · Applied Sciences · doi:10.3390/app10165662 · Cited by 20
Ayad, A., Hallawa, A., Peine, A., Martin, L., Fazlic, L. B., Dartmann, G., Marx, G., & Schmeink, A. (2022) · JMIR Medical Informatics · doi:10.2196/37658 · Cited by 14
Chrominance and Luminance: a study to detect deepfakes
Talib, M. A., Nasir, Q., Nassif, A. B., Fadhl, N. B., & Gouda, O. M. (2025) · Multimedia Tools and Applications · doi:10.1007/s11042-025-20618-3 · Cited by 5
Uncertainty Quantification in Deep Residual Neural Networks
Wandzik, L., Garcia, R. V., & Krüger, J. (2020) · arXiv · doi:10.48550/arxiv.2007.04905 · Cited by 1
Metadata
- Authors
- Li, X., Chen, S., Hu, X., & Yang, J.
- Venue
- 2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR)
- Publisher
- IEEE
- Published
- 2019-06
- Pages
- 2677-2685
- ISSN
- 1063-6919 (ISSN-L)
- OpenAlex
- W2962734576
- Cited by
- 318
- References
- 40
- FWCI
- 16.78
- Type
- conference paper
- Language
- en
- Open access
- green
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.