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

Table 8

The performance of watermarking under pruning when .

SN	Pruning rate (%)	SST-2		AG-News
		Testing acc (%)	WM query	Testing acc (%)	WM query
SN-20-1	0	88.58	Lossless	93.57	Lossless
	90	88.96	Lossless	85.49	[11,11,11,11,11,10,11,11,11,11,11,11,11,11,11,10,11,11,11,11]
SN-20-2	0	88.63	Lossless	93.39	Lossless
	90	87.48	Lossless	84.67	[11,11,11,11,11,11,10,11,11,11,11,11,11,11,11,11,11,11,11,11]
SN-20-10	0	88.69	Lossless	93.46	Lossless
	90	88.66	Lossless	86.97	[11,11,11,11,11,11,10,11,11,11,11,11,11,11,11,11,11,11,11,11]
SN-20-20	0	88.80	Lossless	93.49	Lossless
	90	88.41	Lossless	87.26	Lossless