Ian Goodfellow, Generative Adversarial Networks (2014) 是 GAN 的经典和奠基性论文,使用全连接层进行上采样。

Radford, Metz, and Chintala, Deep Convolutional Generative Adversarial Networks (DCGANs) (2015) 是首次在 GAN 中引入了卷积神经网络,并且引入了转置卷积 (transposed convolutions) 来进行上采样。