stem/AI/Neural Networks/CNN/GAN/GAN.md

# Fully [[Convolution]]al
- Remove [[Max Pooling]]
	- Use strided [[upconv]]
- Remove [[MLP|FC]] layers
	- Hurts convergence in non-classification
- Normalisation tricks
	- Batch normalisation
		- Batches of 0 mean and variance 1
	- Leaky [[Activation Functions#ReLu|ReLu]]

# Stages
## Generator, G
- Synthesise 'fake' images
- From noise
## Discriminator, D
- Discriminator is a classifier
	- Is image fake or real

![[gan-arch.png]]
![[gan-arch2.png]]

![[gan-results.png]]

# Training
![[gan-training-discriminator.png]]
![[gan-training-generator.png]]

# Code Vector Math for Control
![[cvmfc.png]]
- Do [[Interpretation#Activation Maximisation|AM]] to derive code for an image
![[code-vector-math-for-control-results.png]]