Toward Learning Human-aligned Cross-domain Robust Models by Countering Misaligned Features

OOD
Re-assign community