Domain Generalization with MixStyle

Abstract

Though convolutional neural networks (CNNs) have demonstrated remarkable ability in learning discriminative features, they often generalize poorly to unseen domains. Domain generalization aims to address this problem by learning from a set of source domains a model that is generalizable to any unseen domain. In this paper, a novel approach is proposed based on probabilistically mixing instance-level feature statistics of training samples across source domains. Our method, termed MixStyle, is motivated by the observation that visual domain is closely related to image style (e.g., photo vs.~sketch images). Such style information is captured by the bottom layers of a CNN where our proposed style-mixing takes place. Mixing styles of training instances results in novel domains being synthesized implicitly, which increase the domain diversity of the source domains, and hence the generalizability of the trained model. MixStyle fits into mini-batch training perfectly and is extremely easy to implement. The effectiveness of MixStyle is demonstrated on a wide range of tasks including category classification, instance retrieval and reinforcement learning.

ICLR 2021; Code is available at https://github.com/KaiyangZhou/mixstyle-release

References in corpus (21)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Playing Atari with Deep Reinforcement Learning
• Instance Normalization: The Missing Ingredient for Fast Stylization
• Improved Regularization of Convolutional Neural Networks with Cutout
• A Learned Representation For Artistic Style
• CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features
• IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures
• DropBlock: A regularization method for convolutional networks
• Manifold Mixup: Better Representations by Interpolating Hidden States
• Domain Generalization via Model-Agnostic Learning of Semantic Features
• Reinforcement Learning with Augmented Data
• A Study on Overfitting in Deep Reinforcement Learning
• Domain Randomization for Transferring Deep Neural Networks from Simulation to the Real World
• Unlabeled Samples Generated by GAN Improve the Person Re-identification Baseline in vitro
• Performance Measures and a Data Set for Multi-Target, Multi-Camera Tracking
• Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixels
• Illuminating Generalization in Deep Reinforcement Learning through Procedural Level Generation
• Torchreid: A Library for Deep Learning Person Re-Identification in Pytorch
• Generalization and Regularization in DQN
• Generalization in Reinforcement Learning with Selective Noise Injection and Information Bottleneck
• Episodic Training for Domain Generalization

Cited by in corpus (22)

• How Neural Networks Extrapolate: From Feedforward to Graph Neural Networks
• An overview of mixing augmentation methods and augmentation strategies
• Learning Generalisable Omni-Scale Representations for Person Re-Identification
• Learning Debiased Representation via Disentangled Feature Augmentation
• Test-time Batch Statistics Calibration for Covariate Shift
• MixStyle Neural Networks for Domain Generalization and Adaptation
• Semi-Supervised Domain Generalization with Stochastic StyleMatch
• Multiple Domain Experts Collaborative Learning: Multi-Source Domain Generalization For Person Re-Identification
• Learning Domain Invariant Representations by Joint Wasserstein Distance Minimization
• A Bit More Bayesian: Domain-Invariant Learning with Uncertainty
• Ranking Distance Calibration for Cross-Domain Few-Shot Learning
• SWAD: Domain Generalization by Seeking Flat Minima
• Feature Stylization and Domain-aware Contrastive Learning for Domain Generalization
• Global and Local Alignment Networks for Unpaired Image-to-Image Translation
• StyleAugment: Learning Texture De-biased Representations by Style Augmentation without Pre-defined Textures
• Unsupervised Domain Generalization for Person Re-identification: A Domain-specific Adaptive Framework
• A Style and Semantic Memory Mechanism for Domain Generalization
• Collaborative Semantic Aggregation and Calibration for Federated Domain Generalization
• Confounder Identification-free Causal Visual Feature Learning
• Semi-Supervised Domain Generalization with Evolving Intermediate Domain
• Towards Data-Free Domain Generalization
• Domain-Specific Bias Filtering for Single Labeled Domain Generalization