BriefGPT.xyz
Dec, 2020
EC-GAN: 使用半监督算法和 GAN 进行低样本分类
EC-GAN: Low-Sample Classification using Semi-Supervised Algorithms and GANs
HTML
PDF
Ayaan Haque
TL;DR
提出了一种名为External Classifier GAN(EC-GAN)的新型GAN模型,利用GAN和半监督算法来改进完全监督区域的分类。通过将外部分类器附加到GAN的生成器上,我们的实验结果表明EC-GAN的表现可与共享架构方法相媲美,远优于标准的数据增强和基于正则化的方法,且可用于小规模、现实数据集。
Abstract
semi-supervised learning
has been gaining attention as it allows for performing image analysis tasks such as
classification
with limited labeled data. Some popular algorithms using
→