PixelCNN models with auxiliary variables for natural image modeling

A. Kolesnikov, C. Lampert, in:, Omnipress, 2017, pp. 1905–1914.

Conference Paper | Published | English
Department
Series Title
PMLR
Abstract
We study probabilistic models of natural images and extend the autoregressive family of PixelCNN models by incorporating latent variables. Subsequently, we describe two new generative image models that exploit different image transformations as latent variables: a quantized grayscale view of the image or a multi-resolution image pyramid. The proposed models tackle two known shortcomings of existing PixelCNN models: 1) their tendency to focus on low-level image details, while largely ignoring high-level image information, such as object shapes, and 2) their computationally costly procedure for image sampling. We experimentally demonstrate benefits of our LatentPixelCNN models, in particular showing that they produce much more realistically looking image samples than previous state-of-the-art probabilistic models.
Publishing Year
Date Published
2017-01-01
Volume
70
Page
1905 - 1914
Conference
ICML: International Conference on Machine Learning
Conference Location
Sydney, Australia
Conference Date
2017-08-06 – 2017-08-11
IST-REx-ID

Cite this

Kolesnikov A, Lampert C. PixelCNN models with auxiliary variables for natural image modeling. In: Vol 70. Omnipress; 2017:1905-1914.
Kolesnikov, A., & Lampert, C. (2017). PixelCNN models with auxiliary variables for natural image modeling (Vol. 70, pp. 1905–1914). Presented at the ICML: International Conference on Machine Learning, Sydney, Australia: Omnipress.
Kolesnikov, Alexander, and Christoph Lampert. “PixelCNN Models with Auxiliary Variables for Natural Image Modeling,” 70:1905–14. Omnipress, 2017.
A. Kolesnikov and C. Lampert, “PixelCNN models with auxiliary variables for natural image modeling,” presented at the ICML: International Conference on Machine Learning, Sydney, Australia, 2017, vol. 70, pp. 1905–1914.
Kolesnikov A, Lampert C. 2017. PixelCNN models with auxiliary variables for natural image modeling. ICML: International Conference on Machine Learning, PMLR, vol. 70. 1905–1914.
Kolesnikov, Alexander, and Christoph Lampert. PixelCNN Models with Auxiliary Variables for Natural Image Modeling. Vol. 70, Omnipress, 2017, pp. 1905–14.

Link(s) to Main File(s)
Access Level
OA Open Access

Export

Marked Publications

Open Data IST Research Explorer

Search this title in

Google Scholar
ISBN Search