A naukri.com initiative
Arxiv
4d
101
Image Credit: Arxiv
Pseudo Multi-Source Domain Generalization: Bridging the Gap Between Single and Multi-Source Domain Generalization
- Deep learning models face challenges in maintaining performance on data distributions different from their training data.
- Multi-source Domain Generalization (MDG) shows promise but creating multi-domain datasets is difficult and costly.
- To address this, Pseudo Multi-source Domain Generalization (PMDG) framework is proposed.
- PMDG generates pseudo-domains from a single source domain using style transfer and data augmentation, enabling the use of MDG algorithms in Single-source Domain Generalization (SDG) settings.
Read Full Article
6 Likes
For uninterrupted reading, download the app