Reducing Domain Gap by Reducing Style Bias

By Hyeonseob Nam, Hyunjae Lee, Jongchan Park, Wonjun Yoon, Donggeun Yoo. 2021

471 citations recorded by the source corpus.

Convolutional Neural Networks (CNNs) often fail to maintain their performance when they confront new test domains, which is known as the problem of domain shift. Recent studies suggest that one of the main causes of this problem is CNNs’ strong inductive bias towards image styles (i.e. textures) which are sensitive to domain changes, rather than contents (i.e. shapes). Inspired by this, we propose to reduce the intrinsic style bias of CNNs to close the gap between domains. Our Style-Agnostic Networks (SagNets) disentangle style encodings from class categories to prevent style biased predictions and focus more on the contents. Extensive experiments show that our method effectively reduces the style bias and makes the model more robust under domain shift. It achieves remarkable performance improvements in a wide range of cross-domain tasks including domain generalization, unsupervised domain adaptation, and semi-supervised domain adaptation on multiple datasets.1

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Jinkyu Kim
2. Jaekoo Lee
3. Daehee Kim
4. Seunghyun Park

In Computer Vision and Pattern Recognition

1. Jinkyu Kim
2. Jaekoo Lee
3. Daehee Kim
4. Seunghyun Park

Citations in the corpus

• SWAD: Domain Generalization by Seeking Flat Minima (2021, neutral): “ERM [29] 85.5 77.5 66.5 46.1 40.9 63.3 Best SOTA competitor 86.6 [30] 78.8 [31] 68.7 [31] 48.6 [32] 43.6 [15, 33] 65.3 SWAD (proposed) 88.1 79.1 70.6 50.0 46.5 66.9 Previous SOTA [31] + SWAD 88.3 78.9 71.3 51.0 46.8 67.3 12], obtained well-transferable model parameters by the meta-learning…”
• Generalizing to Unseen Domains: A Survey on Domain Generalization (2021, unknown): “[127] proposed to disentangle the style and”
• Single-Source Domain Expansion Network for Cross-Scene Hyperspectral Image Classification (2022, unknown): “alization) [42] and Style-Agnostic Network (SagNet) [43].”
• SelfReg: Self-supervised Contrastive Regularization for Domain Generalization (2021, background): “This testbed currently provides seven multi-domain datasets (i.e. ColoredMNIST [1], RotatedMNIST [16], VLCS [13], PACS [26], Office-Home [42], and TerraIncognita [2], DomainNet [34]) and provides benchmarks results of 14 baseline approaches (i.e. ERM [41], IRM [1], GroupDRO [36], Mixup [45], MLDG [27”