Variational Disentanglement for Domain Generalization

Abstract

Domain generalization aims to learn an invariant model that can generalize well to the unseen target domain. In this paper, we propose to tackle the problem of domain generalization by delivering an effective framework named Variational Disentanglement Network (VDN), which is capable of disentangling the domain-specific features and task-specific features, where the task-specific features are expected to be better generalized to unseen but related test data. We further show the rationale of our proposed method by proving that our proposed framework is equivalent to minimize the evidence upper bound of the divergence between the distribution of task-specific features and its invariant ground truth derived from variational inference. We conduct extensive experiments to verify our method on three benchmarks, and both quantitative and qualitative results illustrate the effectiveness of our method.

arXiv commentAccepted to TMLR 2022

References in corpus (20)

• Unpaired Image-to-Image Translation using Cycle-Consistent Adversarial Networks
• SGDR: Stochastic Gradient Descent with Warm Restarts
• Understanding deep learning requires rethinking generalization
• f-GAN: Training Generative Neural Samplers using Variational Divergence Minimization
• Domain Generalization via Invariant Feature Representation
• Domain Generalization via Model-Agnostic Learning of Semantic Features
• Generalizing to Unseen Domains via Adversarial Data Augmentation
• Generalizing Across Domains via Cross-Gradient Training
• Invariant Risk Minimization
• Domain Agnostic Learning with Disentangled Representations
• Domain Generalization for Medical Imaging Classification with Linear-Dependency Regularization
• Few-Shot Unsupervised Image-to-Image Translation
• Bias and Generalization in Deep Generative Models: An Empirical Study
• Domain Generalization for Object Recognition with Multi-task Autoencoders
• Invariant Risk Minimization Games
• Episodic Training for Domain Generalization
• Learning from Extrinsic and Intrinsic Supervisions for Domain Generalization
• COCO-FUNIT: Few-Shot Unsupervised Image Translation with a Content Conditioned Style Encoder
• Open Domain Generalization with Domain-Augmented Meta-Learning
• Deep Domain-Adversarial Image Generation for Domain Generalisation