Research Article

TADW: Traceable and Anti-detection Dynamic Watermarking of Deep Neural Networks

Table 6

Testing accuracy on clean models and watermarked models under different .

ModelSST-2AG-News
TextCNN (%)TextRNN (%)BERT (%)TextCNN (%)TextRNN (%)BERT (%)

Clean89.0788.1991.4993.8893.3094.25
SN-15-188.5888.1991.1693.4292.8293.89
SN-15-288.6987.8691.1093.4592.8693.93
SN-15-788.6387.7090.9993.5993.0493.99
SN-15-1088.5888.0390.9993.4592.9194.05
SN-20-188.5887.7591.0593.5792.8793.87
SN-20-288.6387.7091.3893.3992.8493.97
SN-20-1088.6988.0890.9993.4692.8094.14
SN-20-2088.8087.7091.3293.4992.8093.87