Image Modeling with Deep Convolutional Gaussian Mixture Models

被引:0
作者
Gepperth, Alexander [1 ]
Pfuelb, Benedikt [1 ]
机构
[1] Fulda Univ Appl Sci, Fulda, Germany
来源
2021 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN) | 2021年
关键词
Deep Learning; Gaussian Mixture Models; Deep Convolutional Gaussian Mixture Models; Stochastic Gradient Descent;
D O I
10.1109/IJCNN52387.2021.9533745
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In this conceptual work, we present Deep Convolutional Gaussian Mixture Models (DCGMMs): a new formulation of deep hierarchical Gaussian Mixture Models (GMMs) that is particularly suitable for describing and generating images. Vanilla (i.e., flat) GMMs require a very large number of components to describe images well, leading to long training times and memory issues. DCGMMs avoid this by a stacked architecture of multiple GMM layers, linked by convolution and pooling operations. This allows to exploit the compositionality of images in a similar way as deep CNNs do. DCGMMs can be trained end-to-end by Stochastic Gradient Descent. This sets them apart from vanilla GMMs which are trained by Expectation-Maximization, requiring a prior k-means initialization which is infeasible in a layered structure. For generating sharp images with DCGMMs, we introduce a new gradient-based technique for sampling through non-invertible operations like convolution and pooling. Based on the MNIST and FashionMNIST datasets, we validate the DCGMMs model by demonstrating its superiority over flat GMMs for clustering, sampling and outlier detection.
引用
收藏
页数:9
相关论文
共 50 条
  • [1] Modeling Multivariate Spray Characteristics with Gaussian Mixture Models
    Wicker, Markus
    Ates, Cihan
    Okraschevski, Max
    Holz, Simon
    Koch, Rainer
    Bauer, Hans-Joerg
    ENERGIES, 2023, 16 (19)
  • [2] Deep LSTM Enhancement for RUL Prediction Using Gaussian Mixture Models
    Sayah, M.
    Guebli, D.
    Noureddine, Z.
    Al Masry, Z.
    AUTOMATIC CONTROL AND COMPUTER SCIENCES, 2021, 55 (01) : 15 - 25
  • [3] Self-Growing Regularized Gaussian Mixture Models for Image Segmentation
    Guan, Tao
    Wang, Hongxia
    Wang, Yan
    PROCEEDINGS OF THE 2012 INTERNATIONAL CONFERENCE ON COMMUNICATION, ELECTRONICS AND AUTOMATION ENGINEERING, 2013, 181 : 577 - 582
  • [4] A robust aerial image registration method using Gaussian mixture models
    Wu, Chuan
    Wang, Yuanyuan
    Karimi, Hamid Reza
    NEUROCOMPUTING, 2014, 144 : 546 - 552
  • [5] Deep LSTM Enhancement for RUL Prediction Using Gaussian Mixture Models
    M. Sayah
    D. Guebli
    Z. Noureddine
    Z. Al Masry
    Automatic Control and Computer Sciences, 2021, 55 : 15 - 25
  • [6] Novel Image Registration Method Using Multiple Gaussian Mixture Models
    Ye, Peng
    Liu, Fang
    PROCEEDINGS OF 2012 2ND INTERNATIONAL CONFERENCE ON COMPUTER SCIENCE AND NETWORK TECHNOLOGY (ICCSNT 2012), 2012, : 2117 - 2120
  • [7] Estimating the number of components in Gaussian mixture models adaptively for medical image
    Xie, Cong-Hua
    Chang, Jin-Yi
    Liu, Yong-Jun
    OPTIK, 2013, 124 (23): : 6216 - 6221
  • [8] q-Gaussian mixture models for image and video semantic indexing
    Inoue, Nakamasa
    Shinoda, Koichi
    JOURNAL OF VISUAL COMMUNICATION AND IMAGE REPRESENTATION, 2013, 24 (08) : 1450 - 1457
  • [9] Hybrid deep convolutional neural models for iris image recognition
    Winston, J. Jenkin
    Hemanth, D. Jude
    Angelopoulou, Anastassia
    Kapetanios, Epaminondas
    MULTIMEDIA TOOLS AND APPLICATIONS, 2022, 81 (07) : 9481 - 9503
  • [10] Hybrid deep convolutional neural models for iris image recognition
    J. Jenkin Winston
    D. Jude Hemanth
    Anastassia Angelopoulou
    Epaminondas Kapetanios
    Multimedia Tools and Applications, 2022, 81 : 9481 - 9503