Main Notations & Acronyms

The main notations used in the paper:

Symbol	Definition
\(n\)	The number of input samples
\(d\)	The dimensionality of input data
\(\mu\)	The mean of a distribution
\(\sigma^2\)	The variance of a distribution
\(\mathcal{N}(\mu, \sigma^2)\)	Gaussian / Normal distribution with mean \(\mu\), variance \(\sigma^2\)
\(L_p\)	\(L_p\) norm used to measure the magnitude of attacks
\(\epsilon\)	The maximum allowable perturbation in adversarial attacks

The main acronyms used in the paper:

Category	Acronym	Definition
Machine Learning	AI	Artificial Intelligence
	CNN	Convolutional Neural Network
	DNN	Deep Neural Network
	GAN	Generative Adversarial Network
	GPT	Generative Pre-Trained Transfomer
	\(k\)-NN	\(k\)-Nearest Neighbor
	LDA	Linear Discriminant Analysis
	ML	Machine Learning
	PCA	Principle Component Analysis
	ReLU	Rectified Linear activation Unit
	SVM	Support Vector Machine
Common Attacks	BIM	Basic Iterative Method
	C&W	Carlini & Wagner Attacks
	FGSM	Fast Gradient Sign Method
	FAB	Fast Adaptive Boundary
	JSMA	Jacobian-based Saliency Map Approach
	ILCM	Iterative Least-likely Class Method
	MIM	Momentum Iterative Method
	PGD	Projected Gradient Descent
	UAP	Universal Adversarial Perturbation
Data Properties	AdvSNR	Adversarial Signal to Noise Ratio
Data Properties	SNR	Signal to Noise Ratio
Others	CV	Computer Vision
	CL	Computer Linguistics
	IS	Information System
	JCR	Journal Citation Reports
	SEC	Security