SelfReg: Self-supervised Contrastive Regularization for Domain Generalization

By Daehee Kim, Seunghyun Park, Jinkyu Kim, Jaekoo Lee. 2021

349 citations recorded by the source corpus.

In general, an experimental environment for deep learning assumes that the training and the test dataset are sampled from the same distribution. However, in real-world situations, a difference in the distribution between two datasets, i.e. domain shift, may occur, which becomes a major factor impeding the generalization performance of the model. The research field to solve this problem is called domain generalization, and it alleviates the domain shift problem by extracting domain-invariant features explicitly or implicitly. In recent studies, contrastive learning-based domain generalization approaches have been proposed and achieved high performance. These approaches require sampling of the negative data pair. However, the performance of contrastive learning fundamentally depends on quality and quantity of negative data pairs. To address this issue, we propose a new regularization method for domain generalization based on contrastive learning, called self-supervised contrastive regularization (SelfReg). The proposed approach use only positive data pairs, thus it resolves various problems caused by negative pair sampling. Moreover, we propose a class-specific domain perturbation layer (CDPL), which makes it possible to effectively apply mixup augmentation even when only positive data pairs are used. The experimental results show that the techniques incorporated by SelfReg contributed to the performance in a compatible manner. In the recent benchmark, DomainBed, the proposed method shows comparable performance to the conventional state-of-the-art alternatives.

Subfields: Artificial Intelligence, Cancer Research, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning

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. Tao Qin
2. Jindong Wang
3. Nan Shao
4. Zhilin Yang
5. Yujun Chen
6. Yulun Du
7. Yidong Ouyang
8. Chang Liu

In Cancer Research

1. Jindong Wang
2. Cuiling Lan

In Computer Vision and Pattern Recognition

1. Tao Qin
2. Nan Shao
3. Jindong Wang
4. Yulun Du
5. Yidong Ouyang
6. Zhilin Yang
7. Chang Liu
8. Cuiling Lan

In Representation & Self-Supervised Learning

1. Tao Qin
2. Yujun Chen
3. Nan Shao
4. Jindong Wang
5. Yulun Du
6. Yidong Ouyang
7. Zhilin Yang
8. Chang Liu

Citations in the corpus

• ZeroPrompt: Scaling Prompt-Based Pretraining to 1, 000 Tasks Improves Zero-Shot Generalization (2022, background): “Previous approaches in domain generalization are mainly based on gradient operations (Gulrajani and Lopez-Paz, 2020; Arjovsky et al., 2019; Mansilla et al., 2021; Kim et al., 2021) or disentanglement learning (Wang et al.”
• Generalizing to Unseen Domains: A Survey on Domain Generalization (2021, methodology): “2 [216, 100, 159, 52, 135, 58, 40, 216, 81, 59] Semi-supervised domain generalization Strain is partially labeled [170, 217] Federated domain generalization Strain cannot broadcast to the server [218, 219, 138] Open domain generalization Ytrain 6= Ytest [54] Unsupervised domain generalization Strain”