论文标题
增强的平衡甘：少数族裔图像生成
Enhanced Balancing GAN: Minority-class Image Generation
论文作者
论文摘要
生成对抗网络（GAN）是最强大的生成模型之一，但始终需要大型且平衡的数据集进行训练。传统的gan不适用于在高度不平衡的数据集中产生少数级图像。提议平衡GAN（BAGAN）来减轻此问题，但是当不同类别的图像看起来相似时，例如花和细胞。在这项工作中，我们提出了一个有监督的自动编码器，该自动编码器具有中间嵌入模型，以分散标记的潜在向量。随着自动编码器初始化的改进，我们还建立了一个具有梯度罚款（Bagan-GP）的啤酒建筑。我们提出的模型克服了原始蝙蝠的不稳定问题，并更快地收敛到高质量的一代。我们的模型在MNIST时尚，CIFAR-10和一个小规模的医疗图像数据集的不平衡尺度版本上实现了高性能。
Generative adversarial networks (GANs) are one of the most powerful generative models, but always require a large and balanced dataset to train. Traditional GANs are not applicable to generate minority-class images in a highly imbalanced dataset. Balancing GAN (BAGAN) is proposed to mitigate this problem, but it is unstable when images in different classes look similar, e.g. flowers and cells. In this work, we propose a supervised autoencoder with an intermediate embedding model to disperse the labeled latent vectors. With the improved autoencoder initialization, we also build an architecture of BAGAN with gradient penalty (BAGAN-GP). Our proposed model overcomes the unstable issue in original BAGAN and converges faster to high quality generations. Our model achieves high performance on the imbalanced scale-down version of MNIST Fashion, CIFAR-10, and one small-scale medical image dataset.
