Adversarial Attacks on Intrusion Detection Systems Using the LSTM Classifier

In this paper, adversarial attacks on machine learning models and their classification are considered. Methods for assessing the resistance of a long short term memory (LSTM) classifier to adversarial attacks. Jacobian based saliency map attack (JSMA) and fast gradient sign method (FGSM) attacks chosen due to the portability of adversarial examples between machine learning models are discussed in detail. An attack of "poisoning" of the LSTM classifier is proposed. Methods of protection against the considered adversarial attacks are formulated.