StyleALAE is a type of adversarial latent autoencoder that uses a StyleGAN based generator. For this the latent space $\mathcal{W}$ plays the same role as the intermediate latent space in StyleGAN. Therefore, the $G$ network becomes the part of StyleGAN depicted on the right side of the Figure. The left side is a novel architecture that we designed to be the encoder $E$. The StyleALAE encoder has Instance Normalization (IN) layers to extract multiscale style information that is combined into a latent code $w$ via a learnable multilinear map.
Source: Adversarial Latent AutoencodersPaper | Code | Results | Date | Stars |
---|
Task | Papers | Share |
---|---|---|
Domain Adaptation | 1 | 33.33% |
Disentanglement | 1 | 33.33% |
Image Generation | 1 | 33.33% |
Component | Type |
|
---|---|---|
Convolution
|
Convolutions | |
Instance Normalization
|
Normalization | |
StyleGAN
|
Generative Models |