A backdoor attack against LSTM-based text classification systems

Jiazhu Dai, Chuanshuai Chen

With the widespread use of deep learning system in many applications, the adversary has strong incentive to explore vulnerabilities of deep neural networks and manipulate them. Bac…