Bi-level Doubly Variational Learning for Energy-based Latent Variable Models - Université de technologie de Troyes Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Bi-level Doubly Variational Learning for Energy-based Latent Variable Models

Ge Kan
  • Fonction : Auteur
Jinhu Lu
  • Fonction : Auteur
Tian Wang
  • Fonction : Auteur
Lei Huang
  • Fonction : Auteur
Guodong Guo
  • Fonction : Auteur

Résumé

Energy-based latent variable models (EBLVMs) are more expressive than conventional energy-based models. However, its potential on visual tasks are limited by its training process based on maximum likelihood estimate that requires sampling from two intractable distributions. In this paper, we propose Bi-level doubly variational learning (BiDVL), which is based on a new bi-level optimization framework and two tractable variational distributions to facilitate learning EBLVMs. Particularly, we lead a decoupled EBLVM consisting of a marginal energy-based distribution and a structural posterior to handle the difficulties when learning deep EBLVMs on images. By choosing a symmetric KL divergence in the lower level of our framework, a compact BiDVL for visual tasks can be obtained. Our model achieves impressive image generation performance over related works. It also demonstrates the significant capacity of testing image reconstruction and out-of-distribution detection.

Dates et versions

hal-04454195 , version 1 (13-02-2024)

Identifiants

Citer

Ge Kan, Jinhu Lu, Tian Wang, Baochang Zhang, Aichun Zhu, et al.. Bi-level Doubly Variational Learning for Energy-based Latent Variable Models. 2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Jun 2022, New Orleans, United States. pp.18439-18448, ⟨10.1109/CVPR52688.2022.01791⟩. ⟨hal-04454195⟩
15 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More