Therefore, the combination of diverse source domains in a single common domain helps to enlarge the data distribution and confirms the effectiveness of SUDA methods in certain cases [ 27 , 41 , 59 ]; however, the improvement might not be significant or guaranteed in many domain-shifting tasks.
← all excerpts
Multi-Feature Unsupervised Domain Adaptation (M-FUDA) Applied to Cross Unaligned Domain-Specific Distributions in Device-Free Human Activity Classification.
1
—
—