AMP

2000년 도입 · 논문 92편에서 사용

Based on the understanding that the flat local minima of the empirical risk cause the model to generalize better. Adversarial Model Perturbation (AMP) improves generalization via minimizing the AMP loss, which is obtained from the empirical risk by applying the worst norm-bounded perturbation on each point in the parameter space.

출처: Regularizing Neural Networks via Adversarial Model Perturbation

소개 논문: Regularizing Neural Networks via Adversarial Model Perturbation