A Deep Generative Deconvolutional Image Model

By Yunchen Pu, Xin Yuan, Andrew Stevens, Chunyuan Li, Lawrence Carin. 2015

45 citations recorded by the source corpus.

A deep generative model is developed for representation and analysis of images, based on a hierarchical convolutional dictionary-learning framework. Stochastic {\em unpooling} is employed to link consecutive layers in the model, yielding top-down image generation. A Bayesian support vector machine is linked to the top-layer features, yielding max-margin discrimination. Deep deconvolutional inference is employed when testing, to infer the latent features, and the top-layer features are connected with the max-margin classifier for discrimination tasks. The model is efficiently trained using a Monte Carlo expectation-maximization (MCEM) algorithm, with implementation on graphical processor units (GPUs) for efficient large-scale learning, and fast testing. Excellent results are obtained on several benchmark datasets, including ImageNet, demonstrating that the proposed model achieves results that are highly competitive with similarly sized convolutional neural networks.

Subfields: Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

1. Chunyuan Li
2. Lawrence Carin
3. Zhe Gan
4. Yunchen Pu
5. Yizhe Zhang
6. Ricardo Henao
7. Jianfeng Gao
8. Luc Van Gool

Citations in the corpus

• Deconvolutional Paragraph Representation Learning (2017, background): “, transposed convolutional) neural network [24, 25] as decoder.”
• Variational Autoencoder for Deep Learning of Images, Labels and Captions (2016, methodology): “The DGDN [8] is used as a decoder, and the encoder for the distribution of latent DGDN parameters is based on a CNN (termed a “recognition model” [10, 11]).”
• Adversarial Symmetric Variational Autoencoder (2017, background): “Generative models are often characterized by latent variables (codes), and the variability in the codes encompasses the variation in the data [1, 2].”
• Collaborative Variational Autoencoder for Recommender Systems (2017, methodology): “Beyond MLP, a deconvolutional network [19, 20] can be applied to model the generation of images, etc.”
• Convolutional Neural Networks Analyzed via Convolutional Sparse Coding (2016, methodology): “Lastly, we exploit the answers to the above questions in order to propose an alternative to the commonly used forward pass algorithm, which is tightly connected to both deconvolutional (Zeiler et al., 2010; Pu et al., 2016) and recurrent networks (Bengio et al.”