Viet-Anh on Software Logo

What is: Group Decreasing Network?

SourceSemantically Multi-modal Image Synthesis
Year2000
Data SourceCC BY-SA - https://paperswithcode.com

Group Decreasing Network, or GroupDNet, is a type of convolutional neural network for multi-modal image synthesis. GroupDNet contains one encoder and one decoder. Inspired by the idea of VAE and SPADE, the encoder EE produces a latent code ZZ that is supposed to follow a Gaussian distribution N(0,1)\mathcal{N}(0,1) during training. While testing, the encoder EE is discarded. A randomly sampled code from the Gaussian distribution substitutes for ZZ. To fulfill this, the re-parameterization trick is used to enable a differentiable loss function during training. Specifically, the encoder predicts a mean vector and a variance vector through two fully connected layers to represent the encoded distribution. The gap between the encoded distribution and Gaussian distribution can be minimized by imposing a KL-divergence loss.