Reproduce “Clarinet: A One-step Approach Towards Budget-friendly Unsupervised Domain Adaptation” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Yiyang Zhang, Feng Liu, Zhen Fang, Bo Yuan, Guangquan Zhang, Jie Lu

CLARINET adapts unlabeled target domains using complementary labels from source examples in one training process. It combines a complementary-label learning objective with adversarial domain alignment and sharpens class predictions before alignment. Across six digit-domain transfer tasks, CLARINET achieves the highest target accuracy on each, averaging 82.426% versus 79.263% for GAC+CDAN. These results do not measure annotation costs or establish gains in human-annotated real-world settings.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.