Watermarking-based Defense against Adversarial Attacks on Deep Neural Networks | IEEE Conference Publication | IEEE Xplore