A Deep Generative Deconvolutional Image Model

Yunchen Pu, Win Yuan, Andrew Stevens, Chunyuan Li, Lawrence Carin
; Proceedings of the 19th International Conference on Artificial Intelligence and Statistics, PMLR 51:741-750, 2016.

Abstract

A deep generative model is developed for representation and analysis of images, based on a hierarchical convolutional dictionary-learning framework. Stochastic 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 algorithm is efficiently trained via Monte Carlo expectation-maximization (MCEM), 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.

Cite this Paper


BibTeX
@InProceedings{pmlr-v51-pu16, title = {A Deep Generative Deconvolutional Image Model}, author = {Yunchen Pu and Win Yuan and Andrew Stevens and Chunyuan Li and Lawrence Carin}, booktitle = {Proceedings of the 19th International Conference on Artificial Intelligence and Statistics}, pages = {741--750}, year = {2016}, editor = {Arthur Gretton and Christian C. Robert}, volume = {51}, series = {Proceedings of Machine Learning Research}, address = {Cadiz, Spain}, month = {09--11 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v51/pu16.pdf}, url = {http://proceedings.mlr.press/v51/pu16.html}, abstract = {A deep generative model is developed for representation and analysis of images, based on a hierarchical convolutional dictionary-learning framework. Stochastic 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 algorithm is efficiently trained via Monte Carlo expectation-maximization (MCEM), 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.} }
Endnote
%0 Conference Paper %T A Deep Generative Deconvolutional Image Model %A Yunchen Pu %A Win Yuan %A Andrew Stevens %A Chunyuan Li %A Lawrence Carin %B Proceedings of the 19th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2016 %E Arthur Gretton %E Christian C. Robert %F pmlr-v51-pu16 %I PMLR %J Proceedings of Machine Learning Research %P 741--750 %U http://proceedings.mlr.press %V 51 %W PMLR %X A deep generative model is developed for representation and analysis of images, based on a hierarchical convolutional dictionary-learning framework. Stochastic 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 algorithm is efficiently trained via Monte Carlo expectation-maximization (MCEM), 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.
RIS
TY - CPAPER TI - A Deep Generative Deconvolutional Image Model AU - Yunchen Pu AU - Win Yuan AU - Andrew Stevens AU - Chunyuan Li AU - Lawrence Carin BT - Proceedings of the 19th International Conference on Artificial Intelligence and Statistics PY - 2016/05/02 DA - 2016/05/02 ED - Arthur Gretton ED - Christian C. Robert ID - pmlr-v51-pu16 PB - PMLR SP - 741 DP - PMLR EP - 750 L1 - http://proceedings.mlr.press/v51/pu16.pdf UR - http://proceedings.mlr.press/v51/pu16.html AB - A deep generative model is developed for representation and analysis of images, based on a hierarchical convolutional dictionary-learning framework. Stochastic 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 algorithm is efficiently trained via Monte Carlo expectation-maximization (MCEM), 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. ER -
APA
Pu, Y., Yuan, W., Stevens, A., Li, C. & Carin, L.. (2016). A Deep Generative Deconvolutional Image Model. Proceedings of the 19th International Conference on Artificial Intelligence and Statistics, in PMLR 51:741-750

Related Material