TL;DR本研究提出一种基于度量空间和开放集 GAN 结构的生成模型,利用度量学习模型提取语义信息来生成类别和细粒度相似的图像,支持生成超越 GAN 训练样本集的图像,数据增强实验表明该方法显著提高了分类器的性能。
Abstract
Many existing conditional generative adversarial networks (cGANs) are limited
to conditioning on pre-defined and fixed class-level semantic labels or
attributes. We propose an open set GAN architecture (OpenGAN) that is
conditioned per-input sample with a feature embedding drawn from a