DC Field | Value | Language |
---|---|---|
dc.contributor.author | Benz, Philipp | ko |
dc.contributor.author | Zhang, Chaoning | ko |
dc.contributor.author | Karjauv, Adil | ko |
dc.contributor.author | Kweon, In So | ko |
dc.date.accessioned | 2021-10-29T06:40:36Z | - |
dc.date.available | 2021-10-29T06:40:36Z | - |
dc.date.created | 2021-10-27 | - |
dc.date.issued | 2021-01 | - |
dc.identifier.citation | IEEE Winter Conference on Applications of Computer Vision (WACV), pp.494 - 503 | - |
dc.identifier.issn | 2472-6737 | - |
dc.identifier.uri | http://hdl.handle.net/10203/288441 | - |
dc.description.abstract | The performance of DNNs trained on clean images has been shown to decrease when the test images have common corruptions. In this work, we interpret corruption robustness as a domain shift and propose to rectify batch normalization (BN) statistics for improving model robustness. This is motivated by perceiving the shift from the clean domain to the corruption domain as a style shift that is represented by the BN statistics. We find that simply estimating and adapting the BN statistics on a few (32 for instance) representation samples, without retraining the model, improves the corruption robustness by a large margin on several benchmark datasets with a wide range of model architectures. For example, on ImageNet-C, statistics adaptation improves the top1 accuracy of ResNet50 from 39.2% to 48.7%. Moreover, we find that this technique can further improve state-of-the-art robust models from 58.1% to 63.3%. | - |
dc.language | English | - |
dc.publisher | IEEE COMPUTER SOC | - |
dc.title | Revisiting Batch Normalization for Improving Corruption Robustness | - |
dc.type | Conference | - |
dc.identifier.wosid | 000692171000050 | - |
dc.identifier.scopusid | 2-s2.0-85098677258 | - |
dc.type.rims | CONF | - |
dc.citation.beginningpage | 494 | - |
dc.citation.endingpage | 503 | - |
dc.citation.publicationname | IEEE Winter Conference on Applications of Computer Vision (WACV) | - |
dc.identifier.conferencecountry | US | - |
dc.identifier.conferencelocation | Waikoloa, HI | - |
dc.identifier.doi | 10.1109/WACV48630.2021.00054 | - |
dc.contributor.localauthor | Kweon, In So | - |
dc.contributor.nonIdAuthor | Benz, Philipp | - |
dc.contributor.nonIdAuthor | Zhang, Chaoning | - |
dc.contributor.nonIdAuthor | Karjauv, Adil | - |
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.