Domain Generalisation with Domain Augmented Supervised Contrastive Learning (Student Abstract)
This work addresses the problem of improving model generalization across different data domains for machine learning practitioners, offering an incremental improvement over existing data augmentation techniques.
This paper tackles the problem of domain shift, where training and target data distributions mismatch. The authors propose a new method combining data augmentation and domain distance minimization, which empirically outperforms baseline results on domain generalization benchmarks.
Domain generalisation (DG) methods address the problem of domain shift, when there is a mismatch between the distributions of training and target domains. Data augmentation approaches have emerged as a promising alternative for DG. However, data augmentation alone is not sufficient to achieve lower generalisation errors. This project proposes a new method that combines data augmentation and domain distance minimisation to address the problems associated with data augmentation and provide a guarantee on the learning performance, under an existing framework. Empirically, our method outperforms baseline results on DG benchmarks.