Stabilizing GANs with Octave Convolutions
- In this preliminary report, we present a simple but very effective technique to stabilize the training of CNN based GANs. Motivated by recently published methods using frequency decomposition of convolutions (e.g. Octave Convolutions), we propose a novel convolution scheme to stabilize the training and reduce the likelihood of a mode collapse. The basic idea of our approach is to splitIn this preliminary report, we present a simple but very effective technique to stabilize the training of CNN based GANs. Motivated by recently published methods using frequency decomposition of convolutions (e.g. Octave Convolutions), we propose a novel convolution scheme to stabilize the training and reduce the likelihood of a mode collapse. The basic idea of our approach is to split convolutional filters into additive high and low frequency parts, while shifting weight updates from low to high during the training. Intuitively, this method forces GANs to learn low frequency coarse image structures before descending into fine (high frequency) details. Our approach is orthogonal and complementary to existing stabilization methods and can simply plugged into any CNN based GAN architecture. First experiments on the CelebA dataset show the effectiveness of the proposed method.…
Document Type: | Article (unreviewed) |
---|---|
Zitierlink: | https://opus.hs-offenburg.de/3958 | Bibliografische Angaben |
Title (English): | Stabilizing GANs with Octave Convolutions |
Author: | Ricard Durall Lopez, Franz-Josef Pfreundt, Janis KeuperStaff MemberORCiDGND |
Year of Publication: | 2019 |
Publisher: | arXiv e-print |
Page Number: | 8 |
URL: | https://onikle.com/articles/13887 |
Language: | English | Inhaltliche Informationen |
Institutes: | Fakultät Elektrotechnik, Medizintechnik und Informatik (EMI) (ab 04/2019) |
Institutes: | Bibliografie |
DDC classes: | 000 Allgemeines, Informatik, Informationswissenschaft |
Tag: | Machine Learning | Formale Angaben |
Open Access: | Open Access |
Licence (German): | Urheberrechtlich geschützt |
Comment: | last revised 17 Dec 2020 |
ArXiv Id: | http://arxiv.org/abs/1905.12534 |