leo-p
commented
7 years ago Summary:
- Dataset: LSUN and ImageNet 1k.
- Objective: Propose a more stable set of architectures for training GAN and show that they learn good representations of images for supervised learning and generative modeling.
Architecture:
Below are the guidelines for making DCGANs.
And here is a sample network:
A tensorflow implementation can be found here along with an online demo.
Results:
Quite interesting especially concerning the structure learned in the Z-space and how this can be used for interpolation or object removal, see the example that is shown everywhere:
Nonetheless the network is still generating small images (32x32).
https://arxiv.org/pdf/1511.06434.pdf
In recent years, supervised learning with convolutional networks (CNNs) has seen huge adoption in computer vision applications. Comparatively, unsupervised learning with CNNs has received less attention. In this work we hope to help bridge the gap between the success of CNNs for supervised learning and unsupervised learning. We introduce a class of CNNs called deep convolutional generative adversarial networks (DCGANs), that have certain architectural constraints, and demonstrate that they are a strong candidate for unsupervised learning. Training on various image datasets, we show convincing evidence that our deep convolutional adversarial pair learns a hierarchy of representations from object parts to scenes in both the generator and discriminator. Additionally, we use the learned features for novel tasks - demonstrating their applicability as general image representations.