논문리뷰(30)
-
[논문리뷰] Defensive Distillation
Adversarial Attack이 있으니 당연히 Defense 기법도 존재한다. 이번 리뷰에서는 Defense 기법 중 하나인 "Defensive Distillation"을 소개할 것이다. 이 논문은 처음 나왔을 때 많은 사람들에게 관심받으며 차세대 defense 기법으로 각광 받았었다. 지금이야 C&W attack이나 PGD 등 여러 Attack이 등장하고 defensive distillation기법이 깨지면서 무뎌졌지만 말이다. https://arxiv.org/abs/1511.04508 Distillation as a Defense to Adversarial Perturbations against Deep Neural Networks Deep learning algorithms have been s..
2022.02.05 -
[논문리뷰] Defense-GAN
이번 리뷰는 Adversarial Examples을 GAN을 사용하여 Defense하는 Defense-GAN에 대한 논문이다. 기존에 GAN에 대해 관심도 가지고 있었는데 이렇게 attack에 GAN을 접목시켜서 Defense를 한다니 흥미롭지 않을 수 없다 ㅎㅎ..살펴보도록 하자! https://arxiv.org/abs/1805.06605 Defense-GAN: Protecting Classifiers Against Adversarial Attacks Using Generative Models In recent years, deep neural network approaches have been widely adopted for machine learning tasks, including classif..
2022.01.28 -
[논문리뷰] StarGAN
이번에는 한국에서 낸 유명한 GAN논문인 StarGAN을 리뷰할 것이다. https://arxiv.org/abs/1711.09020 StarGAN: Unified Generative Adversarial Networks for Multi-Domain Image-to-Image Translation Recent studies have shown remarkable success in image-to-image translation for two domains. However, existing approaches have limited scalability and robustness in handling more than two domains, since different models should be bu..
2022.01.24 -
[논문리뷰] Perceptual Adversarial Networks for Image-to-Image Transformation
https://arxiv.org/abs/1706.09138 Perceptual Adversarial Networks for Image-to-Image Transformation In this paper, we propose a principled Perceptual Adversarial Networks (PAN) for image-to-image transformation tasks. Unlike existing application-specific algorithms, PAN provides a generic framework of learning mapping relationship between paired images ( arxiv.org 최근 GAN에 많은 관심이 있어서 관련 논문을 차근차근 읽..
2022.01.18 -
[논문리뷰] Amplitude-Phase Recombination
요즘 딥러닝과 forier domain 사이의 관계 또는 조합 관련하여 논문을 읽고 있는데, 그 중 흥미로운 논문 하나를 리뷰하려고한다. https://arxiv.org/abs/2108.08487 Amplitude-Phase Recombination: Rethinking Robustness of Convolutional Neural Networks in Frequency Domain Recently, the generalization behavior of Convolutional Neural Networks (CNN) is gradually transparent through explanation techniques with the frequency components decomposition. Howe..
2022.01.12