Source Separation with Deep Generative Priors

被引:0
|
作者
Jayaram, Vivek [1 ]
Thickstun, John [1 ]
机构
[1] Univ Washington, Paul G Allen Sch Comp Sci & Engn, Seattle, WA 98195 USA
基金
美国国家科学基金会;
关键词
BLIND SOURCE SEPARATION;
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
Despite substantial progress in signal source separation, results for richly structured data continue to contain perceptible artifacts. In contrast, recent deep generative models can produce authentic samples in a variety of domains that are indistinguishable from samples of the data distribution. This paper introduces a Bayesian approach to source separation that uses generative models as priors over the components of a mixture of sources, and noise-annealed Langevin dynamics to sample from the posterior distribution of sources given a mixture. This decouples the source separation problem from generative modeling, enabling us to directly use cutting-edge generative models as priors. The method achieves state-of-the-art performance for MNIST digit separation. We introduce new methodology for evaluating separation quality on richer datasets, providing quantitative evaluation of separation results on CIFAR-10. We also provide qualitative results on LSUN.
引用
收藏
页数:12
相关论文
共 50 条
  • [1] Unsupervised Audio Source Separation using Generative Priors
    Narayanaswamy, Vivek
    Thiagarajan, Jayaraman J.
    Anirudh, Rushil
    Spanias, Andreas
    INTERSPEECH 2020, 2020, : 2657 - 2661
  • [2] Underdetermined source separation with structured source priors
    Vincent, E
    Rodet, X
    INDEPENDENT COMPONENT ANALYSIS AND BLIND SIGNAL SEPARATION, 2004, 3195 : 327 - 334
  • [3] Inference with deep generative priors in high dimensions
    Pandit P.
    Sahraee-Ardakan M.
    Rangan S.
    Schniter P.
    Fletcher A.K.
    IEEE. J. Sel. Area. Inf. Theory., 1 (336-347): : 336 - 347
  • [4] Problems using deep generative models for probabilistic audio source separation
    Frank, Maurice
    Ilse, Maximilian
    NEURIPS WORKSHOPS, 2020, 2020, 137 : 53 - 59
  • [5] Robust Compressed Sensing MRI with Deep Generative Priors
    Jalal, Ajil
    Arvinte, Marius
    Daras, Giannis
    Price, Eric
    Dimakis, Alexandros G.
    Tamir, Jonathan I.
    ADVANCES IN NEURAL INFORMATION PROCESSING SYSTEMS 34 (NEURIPS 2021), 2021, 34
  • [6] Deep generative image priors for semantic face manipulation
    Hou, Xianxu
    Shen, Linlin
    Ming, Zhong
    Qiu, Guoping
    PATTERN RECOGNITION, 2023, 139
  • [7] Joint ptycho-tomography with deep generative priors
    Aslan, Selin
    Liu, Zhengchun
    Nikitin, Viktor
    Bicer, Tekin
    Leyffer, Sven
    Gursoy, Doga
    MACHINE LEARNING-SCIENCE AND TECHNOLOGY, 2021, 2 (04):
  • [8] Blind Image Deconvolution Using Deep Generative Priors
    Asim, Muhammad
    Shamshad, Fahad
    Ahmed, Ali
    IEEE TRANSACTIONS ON COMPUTATIONAL IMAGING, 2020, 6 : 1493 - 1506
  • [9] Stable Deep MRI Reconstruction Using Generative Priors
    Zach, Martin
    Knoll, Florian
    Pock, Thomas
    IEEE TRANSACTIONS ON MEDICAL IMAGING, 2023, 42 (12) : 3817 - 3832
  • [10] GENERATIVE ADVERSARIAL SOURCE SEPARATION
    Subakan, Y. Cem
    Smaragdis, Paris
    2018 IEEE INTERNATIONAL CONFERENCE ON ACOUSTICS, SPEECH AND SIGNAL PROCESSING (ICASSP), 2018, : 26 - 30