Data-free distillation from non-transferable teachers fails because synthesized samples drift toward the OOD domain; ATEsc separates ID-like from OOD-like samples via adversarial robustness and improves distillation.
As a result, the generator G is trained to synthesize both ID-like and OOD-like samples (i.e., ID-to-OOD synthetic distribution shift)
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
When Data-Free Knowledge Distillation Meets Non-Transferable Teacher: Escaping Out-of-Distribution Trap is All You Need
Data-free distillation from non-transferable teachers fails because synthesized samples drift toward the OOD domain; ATEsc separates ID-like from OOD-like samples via adversarial robustness and improves distillation.