Deep Convolutional GAN

Introduced by Radford et al. in Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

DCGAN, or Deep Convolutional GAN, is a generative adversarial network architecture. It uses a couple of guidelines, in particular:

Replacing any pooling layers with strided convolutions (discriminator) and fractional-strided convolutions (generator).
Using batchnorm in both the generator and the discriminator.
Removing fully connected hidden layers for deeper architectures.
Using ReLU activation in generator for all layers except for the output, which uses tanh.
Using LeakyReLU activation in the discriminator for all layer.

Source: Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks

Read Paper See Code

Paper	Code	Results	Date	Stars

Task	Papers	Share
Image Generation	20	26.32%
Conditional Image Generation	3	3.95%
Anomaly Detection	2	2.63%
Medical Image Generation	2	2.63%
Object Detection	2	2.63%
Pedestrian Detection	2	2.63%
BIG-bench Machine Learning	2	2.63%
General Classification	2	2.63%
Image Classification	2	2.63%