GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium, Martin Heusel, Hubert Ramsauer, Thomas Unterthiner, Bernhard Nessler, and Sepp Hochreiter, 2017Advances in Neural Information Processing Systems (NeurIPS) 30 (Curran Associates, Inc.) - 介绍了用于稳定GAN训练的双时间尺度更新规则(TTUR)以及用于评估GAN样本质量的Fréchet Inception距离(FID)。
Population Based Training of Neural Networks, Max Jaderberg, Valentin Dalibard, Simon Osindero, Wojciech M. Czarnecki, Jeff Donahue, Ali Razavi, Oriol Vinyals, Tim Green, Iain Dunning, Karen Simonyan, Chrisantha Fernando, Koray Kavukcuoglu, 2017arXiv preprint arXiv:1711.09846DOI: 10.48550/arXiv.1711.09846 - 提出了基于种群的训练(PBT),这是一种通过并行训练一组模型来结合超参数优化和模型训练的方法。
Improved Training of Wasserstein GANs, Ishaan Gulrajani, Faruk Ahmed, Martin Arjovsky, Vincent Dumoulin, Aaron Courville, 2017Advances in Neural Information Processing Systems, Vol. 30 (NeurIPS)DOI: 10.5555/3295222.3295327 - 介绍了WGAN-GP公式,该公式使用梯度惩罚来强制执行Lipschitz约束,显著提高了GAN训练的稳定性。