Conditional Image Synthesis with Auxiliary Classifier GANs

Augustus Odena, Christopher Olah, Jonathon Shlens
Proceedings of the 34th International Conference on Machine Learning, PMLR 70:2642-2651, 2017.

Abstract

In this paper we introduce new methods for the improved training of generative adversarial networks (GANs) for image synthesis. We construct a variant of GANs employing label conditioning that results in $128\times 128$ resolution image samples exhibiting global coherence. We expand on previous work for image quality assessment to provide two new analyses for assessing the discriminability and diversity of samples from class-conditional image synthesis models. These analyses demonstrate that high resolution samples provide class information not present in low resolution samples. Across 1000 ImageNet classes, $128\times 128$ samples are more than twice as discriminable as artificially resized $32\times 32$ samples. In addition, 84.7\% of the classes have samples exhibiting diversity comparable to real ImageNet data.

Cite this Paper

BibTeX
@InProceedings{pmlr-v70-odena17a, title = {Conditional Image Synthesis with Auxiliary Classifier {GAN}s}, author = {Augustus Odena and Christopher Olah and Jonathon Shlens}, booktitle = {Proceedings of the 34th International Conference on Machine Learning}, pages = {2642--2651}, year = {2017}, editor = {Precup, Doina and Teh, Yee Whye}, volume = {70}, series = {Proceedings of Machine Learning Research}, month = {06--11 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v70/odena17a/odena17a.pdf}, url = {https://proceedings.mlr.press/v70/odena17a.html}, abstract = {In this paper we introduce new methods for the improved training of generative adversarial networks (GANs) for image synthesis. We construct a variant of GANs employing label conditioning that results in $128\times 128$ resolution image samples exhibiting global coherence. We expand on previous work for image quality assessment to provide two new analyses for assessing the discriminability and diversity of samples from class-conditional image synthesis models. These analyses demonstrate that high resolution samples provide class information not present in low resolution samples. Across 1000 ImageNet classes, $128\times 128$ samples are more than twice as discriminable as artificially resized $32\times 32$ samples. In addition, 84.7\% of the classes have samples exhibiting diversity comparable to real ImageNet data.} }
Endnote
%0 Conference Paper %T Conditional Image Synthesis with Auxiliary Classifier GANs %A Augustus Odena %A Christopher Olah %A Jonathon Shlens %B Proceedings of the 34th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2017 %E Doina Precup %E Yee Whye Teh %F pmlr-v70-odena17a %I PMLR %P 2642--2651 %U https://proceedings.mlr.press/v70/odena17a.html %V 70 %X In this paper we introduce new methods for the improved training of generative adversarial networks (GANs) for image synthesis. We construct a variant of GANs employing label conditioning that results in $128\times 128$ resolution image samples exhibiting global coherence. We expand on previous work for image quality assessment to provide two new analyses for assessing the discriminability and diversity of samples from class-conditional image synthesis models. These analyses demonstrate that high resolution samples provide class information not present in low resolution samples. Across 1000 ImageNet classes, $128\times 128$ samples are more than twice as discriminable as artificially resized $32\times 32$ samples. In addition, 84.7\% of the classes have samples exhibiting diversity comparable to real ImageNet data.
APA
Odena, A., Olah, C. & Shlens, J.. (2017). Conditional Image Synthesis with Auxiliary Classifier GANs. Proceedings of the 34th International Conference on Machine Learning, in Proceedings of Machine Learning Research 70:2642-2651 Available from https://proceedings.mlr.press/v70/odena17a.html.

Related Material