Tim Salimans, Han Zhang, Alec Radford, Dimitris Metaxas
TL;DR本文介绍了新兴的生成对抗网络--优化输运生成对抗网络,提升了图像生成模型的稳定性和生成效果,主要特点是利用adversarially learned feature space来定义mini-batch energy distance度量优化目标,在多个经典image generation benchmark test中达到了最好的性能表现。
Abstract
We present optimal transport gan (OT-GAN), a variant of generative adversarial nets minimizing a new metric measuring the distance between the generator distribution and the data distribution. This metric, which