Meta-Learning Divergences for Variational Inference
Ruqi Zhang, Yingzhen Li, Christopher De Sa, Sam Devlin, Cheng Zhang
Abstract
Variational inference (VI) plays an essential role in approximate Bayesian inference due to its computational efficiency and broad applicability. Crucial to the performance of VI is the selection of the associated divergence measure, as VI approximates the intractable distribution by minimizing this divergence. In this paper we propose a meta-learning algorithm to learn the divergence metric suited for the task of interest, automating the design of VI methods. In addition, we learn the initialization of the variational parameters without additional cost when our method is deployed in the few-shot learning scenarios. We demonstrate our approach outperforms standard VI on Gaussian mixture distribution approximation, Bayesian neural network regression, image generation with variational autoencoders and recommender systems with a partial variational autoencoder.
BibTeX
@InProceedings{pmlr-v130-zhang21o,
title = { Meta-Learning Divergences for Variational Inference },
author = {Zhang, Ruqi and Li, Yingzhen and De Sa, Christopher and Devlin, Sam and Zhang, Cheng},
booktitle = {Proceedings of The 24th International Conference on Artificial Intelligence and Statistics},
pages = {4024--4032},
year = {2021},
editor = {Banerjee, Arindam and Fukumizu, Kenji},
volume = {130},
series = {Proceedings of Machine Learning Research},
month = {13--15 Apr},
publisher = {PMLR},
pdf = {http://proceedings.mlr.press/v130/zhang21o/zhang21o.pdf},
url = {https://proceedings.mlr.press/v130/zhang21o.html},
abstract = { Variational inference (VI) plays an essential role in approximate Bayesian inference due to its computational efficiency and broad applicability. Crucial to the performance of VI is the selection of the associated divergence measure, as VI approximates the intractable distribution by minimizing this divergence. In this paper we propose a meta-learning algorithm to learn the divergence metric suited for the task of interest, automating the design of VI methods. In addition, we learn the initialization of the variational parameters without additional cost when our method is deployed in the few-shot learning scenarios. We demonstrate our approach outperforms standard VI on Gaussian mixture distribution approximation, Bayesian neural network regression, image generation with variational autoencoders and recommender systems with a partial variational autoencoder. }
}