A Robust Countermeasures for Poisoning Attacks on Deep Neural Networks of Computer Interaction Systems

In recent years, human–computer interactions have begun to apply deep neural networks (DNNs), known as learning, make them work more friendly. Nowadays, adversarial example attacks, poisoning and backdoor attacks are the typical attack examples for DNNs. this paper, we focus on analyze three We develop a countermeasure which is Data Washing, an algorithm based denoising autoencoder. It can effectively alleviate damages inflicted upon datasets caused by attacks. Furthermore, also propose Integrated Detection Algorithm (IDA) detect various types of our experiments, Paralysis Attacks, Washing represents significant improvement (0.5384) over accuracy increment, help IDA those while Target makes it so that false positive rate reduced just 1% high detection greater than 99%.