2020/07(4)
-
[Paper review] StyleGAN2
Analyzing and Improving the Image Quality of StyleGAN Introduction 제목: Analyzing and Improving the Image Quality of StyleGAN 저자: Tero Karras, Samuli Laine, Miika Aittala, Janne Hellsten, Jaakko Lehtinen, Timo Aila 기관: NVIDIA 요약: StyleGAN의 문제를 분석하고, 모델의 구조와 학습 방법을 개선 Generator의 구조 개선 Redesign generator normalization: "Droplet artifacts" 문제 해결 Revisit progressive growing: "phase" artifacts, Shift-..
2020.07.28 -
[Paper Review] StyleGAN
StyleGAN: A Style-Based Generator Architecture for GANs We propose an alternative generator architecture for generative adversarial networks, borrowing from style transfer literature. The new architecture leads to an automatically learned, unsupervised separation of high-level attributes (e.g., pose and identity when trained on human faces) and stochastic variation in the generated images (e.g.,..
2020.07.27 -
[GAN] 반드시 읽어야 할 논문 리스트
Lists of Must-Read Papers Goodfellow, Ian, et al. “Generative adversarial nets” Advances in neural information processing systems. 2014. Conditional GAN pix2pix CycleGAN Salimans, Tim, et al. “Improved techniques for training gans” Advances in neural information processing systems. 2016. WassersteinGAN Non saturating GAN LSGAN DCGAN Semi-supervised GAN Spectral Normalization for Generative Adver..
2020.07.25 -
Small Batch Size in Deep Learning
How to Choose the Batch Size? 일반적으로 아래 세 가지 batch size를 결정할 수 있으며, 배치 사이즈의 크기에 따라 딥러닝의 성능에 많은 영향을 미친다. Batch(deterministic) Gradient Descent Mini-Batch Gradient Descent Stochastic(Online) Gradient Descent Large Batch Small Batch Accurate estimate of the gradient (low variance) Noisy estimate of the gradient (high variance) High Computation cost per iteration Low computation cost per iteration Hi..
2020.07.23