MixStyle Neural Networks for Domain Generalization and Adaptation

Abstract

Neural networks do not generalize well to unseen data with domain shifts -- a longstanding problem in machine learning and AI. To overcome the problem, we propose MixStyle, a simple plug-and-play, parameter-free module that can improve domain generalization performance without the need to collect more data or increase model capacity. The design of MixStyle is simple: it mixes the feature statistics of two random instances in a single forward pass during training. The idea is grounded by the finding from recent style transfer research that feature statistics capture image style information, which essentially defines visual domains. Therefore, mixing feature statistics can be seen as an efficient way to synthesize new domains in the feature space, thus achieving data augmentation. MixStyle is easy to implement with a few lines of code, does not require modification to training objectives, and can fit a variety of learning paradigms including supervised domain generalization, semi-supervised domain generalization, and unsupervised domain adaptation. Our experiments show that MixStyle can significantly boost out-of-distribution generalization performance across a wide range of tasks including image recognition, instance retrieval and reinforcement learning.

Extension of https://openreview.net/forum?id=6xHJ37MVxxp. Code available at https://github.com/KaiyangZhou/mixstyle-release. To appear in IJCV

References in corpus (10)

• Learning Transferable Features with Deep Adaptation Networks
• Improved Regularization of Convolutional Neural Networks with Cutout
• A Learned Representation For Artistic Style
• Domain Generalization via Invariant Feature Representation
• VisDA: The Visual Domain Adaptation Challenge
• Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World
• Torchreid: A Library for Deep Learning Person Re-Identification in Pytorch
• Generalization in Reinforcement Learning with Selective Noise Injection and Information Bottleneck
• Domain Generalization with MixStyle
• Semi-Supervised Domain Generalization with Stochastic StyleMatch