Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
University of Southampton
3
0
4
Pairing observations in domain adaptation can drastically reduce gradient variance and boost accuracy, outperforming traditional methods.
ARROW achieves online variance reduction for domain adaptation, outperforming traditional offline methods in dynamic data environments.
Training data order matters more than you think: reordering your data can significantly improve unsupervised domain adaptation by reducing variance in domain discrepancy estimates.