DDSR adaptively fuses black-box and ViL predictions with subnetwork rectification and self-training to outperform prior black-box domain adaptation methods on benchmarks.
Clip-adapter: Better vision-language models with feature adapters,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
Adaptive Dual-Teacher Distillation with Subnetwork Rectification for Bridging Semantic Gaps in Black-Box Domain Adaptation
DDSR adaptively fuses black-box and ViL predictions with subnetwork rectification and self-training to outperform prior black-box domain adaptation methods on benchmarks.