Open Domain Generalization with Domain-Augmented Meta-Learning
This addresses a practical challenge in computer vision for real-world applications where annotated data from target domains are unavailable, though it is incremental as it builds on meta-learning and domain generalization techniques.
The paper tackles the problem of Open Domain Generalization (OpenDG), where models must generalize to unseen domains with different distributions and label sets, by proposing a Domain-Augmented Meta-Learning framework that uses Dirichlet mixup and distilled soft-labeling for domain augmentation, achieving improved performance on unseen domain recognition in experiments.
Leveraging datasets available to learn a model with high generalization ability to unseen domains is important for computer vision, especially when the unseen domain's annotated data are unavailable. We study a novel and practical problem of Open Domain Generalization (OpenDG), which learns from different source domains to achieve high performance on an unknown target domain, where the distributions and label sets of each individual source domain and the target domain can be different. The problem can be generally applied to diverse source domains and widely applicable to real-world applications. We propose a Domain-Augmented Meta-Learning framework to learn open-domain generalizable representations. We augment domains on both feature-level by a new Dirichlet mixup and label-level by distilled soft-labeling, which complements each domain with missing classes and other domain knowledge. We conduct meta-learning over domains by designing new meta-learning tasks and losses to preserve domain unique knowledge and generalize knowledge across domains simultaneously. Experiment results on various multi-domain datasets demonstrate that the proposed Domain-Augmented Meta-Learning (DAML) outperforms prior methods for unseen domain recognition.