Linked Presentation: Rethinking White-Box Watermarks on Deep Learning Models under Neural Structural Obfuscation
