Semi-Supervised Learning algorithms can be applied out-of-the-box for Domain Adaptation! This video explains the extensions to FixMatch in AdaMatch and how it has been applied to the DomainNet benchmark for source to target domain transfer!
Paper Links:
AdaMatch:
FixMatch:
DomainNet Project Page:
Learning a Universal Template for Few-Shot Dataset Generalization:
Training BatchNorm and only BatchNorm:
Thanks for watching! Please Subscribe!
Chapters
0:00 Paper Title
0:10 SSL, UDA, and SSDA
2:12 DomainNet Benchmark
4:27 AdaMatch overview
5:11 Technical Details
15:07 Flow of Loss Functions
17:17 Results
19:45 Problem Summary
20:43 BatchNorm for Adaptation