English

Training Energy-Based Normalizing Flow with Score-Matching Objectives

Machine Learning 2023-10-31 v2 Machine Learning

Abstract

In this paper, we establish a connection between the parameterization of flow-based and energy-based generative models, and present a new flow-based modeling approach called energy-based normalizing flow (EBFlow). We demonstrate that by optimizing EBFlow with score-matching objectives, the computation of Jacobian determinants for linear transformations can be entirely bypassed. This feature enables the use of arbitrary linear layers in the construction of flow-based models without increasing the computational time complexity of each training iteration from O(D2L)O(D^2L) to O(D3L)O(D^3L) for an LL-layered model that accepts DD-dimensional inputs. This makes the training of EBFlow more efficient than the commonly-adopted maximum likelihood training method. In addition to the reduction in runtime, we enhance the training stability and empirical performance of EBFlow through a number of techniques developed based on our analysis of the score-matching methods. The experimental results demonstrate that our approach achieves a significant speedup compared to maximum likelihood estimation while outperforming prior methods with a noticeable margin in terms of negative log-likelihood (NLL).

Keywords

Cite

@article{arxiv.2305.15267,
  title  = {Training Energy-Based Normalizing Flow with Score-Matching Objectives},
  author = {Chen-Hao Chao and Wei-Fang Sun and Yen-Chang Hsu and Zsolt Kira and Chun-Yi Lee},
  journal= {arXiv preprint arXiv:2305.15267},
  year   = {2023}
}

Comments

Published at NeurIPS 2023. Code: https://github.com/chen-hao-chao/ebflow

R2 v1 2026-06-28T10:44:46.863Z