Pixel Recurrent Neural Networks

Aaron Van Oord, Nal Kalchbrenner, Koray Kavukcuoglu
; Proceedings of The 33rd International Conference on Machine Learning, PMLR 48:1747-1756, 2016.

Abstract

Modeling the distribution of natural images is a landmark problem in unsupervised learning. This task requires an image model that is at once expressive, tractable and scalable. We present a deep neural network that sequentially predicts the pixels in an image along the two spatial dimensions. Our method models the discrete probability of the raw pixel values and encodes the complete set of dependencies in the image. Architectural novelties include fast two-dimensional recurrent layers and an effective use of residual connections in deep recurrent networks. We achieve log-likelihood scores on natural images that are considerably better than the previous state of the art. Our main results also provide benchmarks on the diverse ImageNet dataset. Samples generated from the model appear crisp, varied and globally coherent.

Cite this Paper


BibTeX
@InProceedings{pmlr-v48-oord16, title = {Pixel Recurrent Neural Networks}, author = {Aaron Van Oord and Nal Kalchbrenner and Koray Kavukcuoglu}, pages = {1747--1756}, year = {2016}, editor = {Maria Florina Balcan and Kilian Q. Weinberger}, volume = {48}, series = {Proceedings of Machine Learning Research}, address = {New York, New York, USA}, month = {20--22 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v48/oord16.pdf}, url = {http://proceedings.mlr.press/v48/oord16.html}, abstract = {Modeling the distribution of natural images is a landmark problem in unsupervised learning. This task requires an image model that is at once expressive, tractable and scalable. We present a deep neural network that sequentially predicts the pixels in an image along the two spatial dimensions. Our method models the discrete probability of the raw pixel values and encodes the complete set of dependencies in the image. Architectural novelties include fast two-dimensional recurrent layers and an effective use of residual connections in deep recurrent networks. We achieve log-likelihood scores on natural images that are considerably better than the previous state of the art. Our main results also provide benchmarks on the diverse ImageNet dataset. Samples generated from the model appear crisp, varied and globally coherent.} }
Endnote
%0 Conference Paper %T Pixel Recurrent Neural Networks %A Aaron Van Oord %A Nal Kalchbrenner %A Koray Kavukcuoglu %B Proceedings of The 33rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2016 %E Maria Florina Balcan %E Kilian Q. Weinberger %F pmlr-v48-oord16 %I PMLR %J Proceedings of Machine Learning Research %P 1747--1756 %U http://proceedings.mlr.press %V 48 %W PMLR %X Modeling the distribution of natural images is a landmark problem in unsupervised learning. This task requires an image model that is at once expressive, tractable and scalable. We present a deep neural network that sequentially predicts the pixels in an image along the two spatial dimensions. Our method models the discrete probability of the raw pixel values and encodes the complete set of dependencies in the image. Architectural novelties include fast two-dimensional recurrent layers and an effective use of residual connections in deep recurrent networks. We achieve log-likelihood scores on natural images that are considerably better than the previous state of the art. Our main results also provide benchmarks on the diverse ImageNet dataset. Samples generated from the model appear crisp, varied and globally coherent.
RIS
TY - CPAPER TI - Pixel Recurrent Neural Networks AU - Aaron Van Oord AU - Nal Kalchbrenner AU - Koray Kavukcuoglu BT - Proceedings of The 33rd International Conference on Machine Learning PY - 2016/06/11 DA - 2016/06/11 ED - Maria Florina Balcan ED - Kilian Q. Weinberger ID - pmlr-v48-oord16 PB - PMLR SP - 1747 DP - PMLR EP - 1756 L1 - http://proceedings.mlr.press/v48/oord16.pdf UR - http://proceedings.mlr.press/v48/oord16.html AB - Modeling the distribution of natural images is a landmark problem in unsupervised learning. This task requires an image model that is at once expressive, tractable and scalable. We present a deep neural network that sequentially predicts the pixels in an image along the two spatial dimensions. Our method models the discrete probability of the raw pixel values and encodes the complete set of dependencies in the image. Architectural novelties include fast two-dimensional recurrent layers and an effective use of residual connections in deep recurrent networks. We achieve log-likelihood scores on natural images that are considerably better than the previous state of the art. Our main results also provide benchmarks on the diverse ImageNet dataset. Samples generated from the model appear crisp, varied and globally coherent. ER -
APA
Oord, A.V., Kalchbrenner, N. & Kavukcuoglu, K.. (2016). Pixel Recurrent Neural Networks. Proceedings of The 33rd International Conference on Machine Learning, in PMLR 48:1747-1756

Related Material