spectral methods of moments provide a powerful tool for learning the
parameters of latent variable models. Despite their theoretical appeal, the
applicability of these methods to real data is still limited due to
本论文提出了一种 Moment Estimation 的算法来训练规模大的 Implicit Generative Models,即 Method of Learned Moments (MoLM)。通过引入 Moment Network,以及使用渐近理论来确定 Moment Estimation 中需要优化的关键性质,MoLM 可以训练出高质量的神经图像生成模型。