Adaptive Multi-stage Density Ratio Estimation for Learning Latent Space Energy-based Model

被引:0
|
作者
Xiao, Zhisheng [1 ]
Han, Tian [2 ]
机构
[1] Univ Chicago, Chicago, IL 60637 USA
[2] Stevens Inst Technol, Hoboken, NJ 07030 USA
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
This paper studies the fundamental problem of learning energy-based model (EBM) in the latent space of the generator model. Learning such prior model typically requires running costly Markov Chain Monte Carlo (MCMC). Instead, we propose to use noise contrastive estimation (NCE) to discriminatively learn the EBM through density ratio estimation between the latent prior density and latent posterior density. However, the NCE typically fails to accurately estimate such density ratio given large gap between two densities. To effectively tackle this issue and learn more expressive prior models, we develop the adaptive multi-stage density ratio estimation which breaks the estimation into multiple stages and learn different stages of density ratio sequentially and adaptively. The latent prior model can be gradually learned using ratio estimated in previous stage so that the final latent space EBM prior can be naturally formed by product of ratios in different stages. The proposed method enables informative and much sharper prior than existing baselines, and can be trained efficiently. Our experiments demonstrate strong performances in image generation and reconstruction as well as anomaly detection.
引用
收藏
页数:12
相关论文
共 50 条
  • [1] Learning Latent Space Energy-Based Prior Model
    Pang, Bo
    Han, Tian
    Nijkamp, Erik
    Zhu, Song-Chun
    Wu, Ying Nian
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 33, NEURIPS 2020, 2020, 33
  • [2] Learning Hierarchical Features with Joint Latent Space Energy-Based Prior
    Cui, Jiali
    Wu, Ying Nian
    Han, Tian
    2023 IEEE/CVF INTERNATIONAL CONFERENCE ON COMPUTER VISION, ICCV, 2023, : 2218 - 2227
  • [3] Learning Generative Vision Transformer with Energy-Based Latent Space for Saliency Prediction
    Zhang, Jing
    Xie, Jianwen
    Barnes, Nick
    Li, Ping
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 34 (NEURIPS 2021), 2021, 34
  • [4] Energy-based Latent Aligner for Incremental Learning
    Joseph, K. J.
    Khan, Salman
    Khan, Fahad Shahbaz
    Anwer, Rao Muhammad
    Balasubramanian, Vineeth N.
    2022 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2022, : 7442 - 7451
  • [6] Cooperative Learning of Energy-Based Model and Latent Variable Model via MCMC Teaching
    Xie, Jianwen
    Lu, Yang
    Gao, Ruiqi
    Wu, Ying Nian
    THIRTY-SECOND AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE / THIRTIETH INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE / EIGHTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2018, : 4292 - 4301
  • [7] Estimation of the odds ratio from multi-stage randomized trials
    Cao, Shiwei
    Jung, Sin-Ho
    PHARMACEUTICAL STATISTICS, 2024,
  • [8] Multi-Stage Feature Constraints Learning for Age Estimation
    Xia, Min
    Zhang, Xu
    Liu, Wan'an
    Weng, Liguo
    Xu, Yiqing
    IEEE TRANSACTIONS ON INFORMATION FORENSICS AND SECURITY, 2020, 15 : 2417 - 2428
  • [9] LEARNING TO DRAW THROUGH A MULTI-STAGE ENVIRONMENT MODEL BASED REINFORCEMENT LEARNING
    Qiu, Ji
    Lu, Peng
    Peng, Xujun
    2023 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING, ICIP, 2023, : 1240 - 1244
  • [10] Multi-stage adaptive enrichment trial design with subgroup estimation
    Joshi, Neha
    Nguyen, Crystal
    Ivanova, Anastasia
    JOURNAL OF BIOPHARMACEUTICAL STATISTICS, 2020, 30 (06) : 1038 - 1049