… ‘consistency’ regularization loss, as a simpler and easy-to-use alternative for regularizing AT. … an auxiliary consistency regularization loss during AT: it forces adversarial examples from …
… training, a method to train a network with iter FGSM adversarial images … adversarial training regularized with a unified embedding for classification and low- level similarity learning by …
C Lyu, K Huang, HN Liang - 2015 IEEE international …, 2015 - ieeexplore.ieee.org
… machinelearning models against adversarial examples. More specifically, using the unified framework, we develop a family of gradient regularization … to deal with adversarial examples. …
… Regularization is a fundamental and incisive method in optimization, its present zeitgeist … machinelearning. Through the introduction of a new component in the objective, regularization …
… The fundamental way to learn a generative model in machinelearning is to (i) define a parametric family of probability densities {Q✓}, ✓ 2 Θ ✓ Rd, and (ii) find parameters ✓⇤ 2 Θ such …
Y Wen, S Li, K Jia - … Conference on Machine Learning, 2020 - proceedings.mlr.press
… The problem of adversarial examples has shown that … In this work, we study the degradation through the regularization … NNs in a gentler way to avoid the problematic regularization. …
… a new regularization method based on virtual adversarial loss: … Virtual adversarial loss is defined as the robustness of the … include simple and scalable machinelearning algorithms. …
M Sato, J Suzuki, S Kiyono - … of the 57th Annual Meeting of the …, 2019 - aclanthology.org
… and benefit of adversarialregularization based on adversarial … Additionally, we confirmed that adversarialregularization … We believe that adversarialregularization can be one of the …
… Abstract—As we seek to deploy machinelearning models beyond … adversarial (re)Training as their main defense against perturbations. We also survey mothods that add regularization …