A Black-Box Attack Method against Machine-Learning-Based Anomaly Network Flow Detection Models

Table 1

Notations and terminology used in this paper.

Notations and symbols	Description
	The original data
	The class that is labeled by the machine learning model
	The adversarial example
	The label of the adversarial examples
	The loss function
	The noise that is added to the original data
	The parameters of the machine learning model
	The target machine learning model or deep learning model
	The substitute model
	The constraint vector of the perturbation
	The influence coefficients vector of the features
	The step size of the perturbation in the single iteration