Normalizing flow neural networks by JKO scheme

arXiv:2212.14424v4 Announce Type: replace-cross
Abstract: Normalizing flow is a class of deep generative models for efficient sampling and likelihood estimation, which achieves attractive performance, particularly in high dimensions. The flow is often implemented using a sequence of invertible residual blocks. Existing works adopt special network architectures and regularization of flow trajectories. In this paper, we develop a neural ODE flow network called JKO-iFlow, inspired by the Jordan-Kinderleherer-Otto (JKO) scheme, which unfolds the discrete-time dynamic of the Wasserstein gradient flow. The …

abstract architectures arxiv class cs.lg deep generative models dimensions flow generative generative models likelihood network networks neural networks performance regularization residual sampling stat.ml type