방법 비교

	적대적 훈련	생성적 적대 신경망	분포 외 탐지
분야	딥러닝	딥러닝	머신러닝
계열	Machine learning	Machine learning	Machine learning
기원 연도	2018	2014	2017
창시자	Aleksander Madry et al.	Goodfellow, I. et al.	Hendrycks & Gimpel
유형	Robust optimization training procedure	Generative deep learning (adversarial two-network game)	Reliability and safety method for neural networks
원전	Madry, A., Makelov, A., Schmidt, L., Tsipras, D., & Vladu, A. (2018). Towards deep learning models resistant to adversarial attacks. International Conference on Learning Representations (ICLR).	Goodfellow, I. et al. (2014). Generative Adversarial Nets. NeurIPS.	Hendrycks, D., & Gimpel, K. (2017). A baseline for detecting misclassified and out-of-distribution examples in neural networks. International Conference on Learning Representations.
별칭	Min-Max Robust Training, PGD Adversarial Training, Robust Empirical Risk Minimization, Hasımsal Eğitim	Üretici Çekişmeli Ağ (GAN), GAN, generative adversarial nets, adversarial network	OOD Detection, Novelty Detection, Open-Set Recognition, Dağılım Dışı Tespit
관련	3	4	3
요약	Adversarial Training is a robust optimization procedure for deep neural networks in which the model is trained not on clean data alone but on worst-case perturbed inputs crafted during training. Formalized by Madry et al. (2018) as a min-max saddle-point problem, the method uses Projected Gradient Descent (PGD) to generate strong adversarial examples within a bounded Lp perturbation set before each gradient update, forcing the network to learn decision boundaries that are stable under such perturbations.	A Generative Adversarial Network (GAN), introduced by Ian Goodfellow and colleagues in 2014, produces realistic synthetic data through the competition of two neural networks — a generator and a discriminator. It is widely used for image synthesis, data augmentation, and distribution estimation.	Out-of-Distribution (OOD) detection is a set of techniques that identify when a deployed machine learning model receives inputs that differ significantly from its training data distribution. Introduced as a formal problem by Hendrycks and Gimpel in 2017, these methods enable models to flag unfamiliar inputs rather than silently produce unreliable predictions, making them foundational to trustworthy and safe AI deployment in high-stakes domains.