id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
cord-024515-iioqkydg	Zhong, Qi	Protecting IP of Deep Neural Networks with Watermarking: A New Label Helps	2020-04-17		.txt	text/plain	4588	233	57	To mitigate this threat, in this paper, we propose an innovative framework to protect the intellectual property of deep learning models, that is, watermarking the model by adding a new label to crafted key samples during training. The intuition comes from the fact that, compared with existing DNN watermarking methods, adding a new label will not twist the original decision boundary but can help the model learn the features of key samples better. Extensive experimental results show that, compared with the existing schemes, the proposed method performs better under small perturbation strength or short key samples' length in terms of classification accuracy and ownership verification efficiency. -Effectiveness and efficiency: the false positive rate for key samples should be minimized, and a reliable ownership verification result needs to be obtained with few queries to the remote DNN API; -Robustness: the watermarked model can resist several known attacks, for example, pruning attack and fine-tuning attack.	./cache/cord-024515-iioqkydg.txt	./txt/cord-024515-iioqkydg.txt