ICML 2022oral18 citations
Path-Gradient Estimators for Continuous Normalizing Flows
Lorenz Vaitl, Kim Andrea Nicoli, Shinichi Nakajima, Pan Kessel
Abstract
Recent work has established a path-gradient estimator for simple variational Gaussian distributions and has argued that the path-gradient is particularly beneficial in the regime in which the variational distribution approaches the exact target distribution. In many applications, this regime can however not be reached by a simple Gaussian variational distribution. In this work, we overcome this crucial limitation by proposing a path-gradient estimator for the considerably more expressive variational family of continuous normalizing flows. We outline an efficient algorithm to calculate this estimator and establish its superior performance empirically.
BibTeX
@InProceedings{pmlr-v162-vaitl22a,
title = {Path-Gradient Estimators for Continuous Normalizing Flows},
author = {Vaitl, Lorenz and Nicoli, Kim Andrea and Nakajima, Shinichi and Kessel, Pan},
booktitle = {Proceedings of the 39th International Conference on Machine Learning},
pages = {21945--21959},
year = {2022},
editor = {Chaudhuri, Kamalika and Jegelka, Stefanie and Song, Le and Szepesvari, Csaba and Niu, Gang and Sabato, Sivan},
volume = {162},
series = {Proceedings of Machine Learning Research},
month = {17--23 Jul},
publisher = {PMLR},
pdf = {https://proceedings.mlr.press/v162/vaitl22a/vaitl22a.pdf},
url = {https://proceedings.mlr.press/v162/vaitl22a.html},
abstract = {Recent work has established a path-gradient estimator for simple variational Gaussian distributions and has argued that the path-gradient is particularly beneficial in the regime in which the variational distribution approaches the exact target distribution. In many applications, this regime can however not be reached by a simple Gaussian variational distribution. In this work, we overcome this crucial limitation by proposing a path-gradient estimator for the considerably more expressive variational family of continuous normalizing flows. We outline an efficient algorithm to calculate this estimator and establish its superior performance empirically.}
}