updates to VAE practicals

deep_learning/README.md

+# Deep Learning Practicals
+
+This folder contains introductions to topics in deep learning. Currently it covers:
+
+- GANs : Generative Adversarial Networks
+- VAEs : Variational Autoencoders
+- RNNs : Recurrent Neural Networks
+
+Pick a folder and run the jupter notebooks within.
+
+

deep_learning/VAEs/README.md

+# Variational AutoEncoders
+
+The notebooks cover:
+
+- `Part 1`: Simple implementation of PCA and Autoencoders (not variationals) in a simple 1D-to-1D example. you will learn the similarities and differences between PCA (a linear auto-encoder) and AEs (which can be nonlinear) and you will be able to easily understand their behaviour because of the siimplicity of the input and output.
+- `Part 2`: Application of Auto-encoders to encoding images (the MNIST dataset), where you will learn how to use AEs for multi-dimensional input and why vanilla AEs do not give good low-dimensional embeddings.
+- `Part 3`: Introducing the variational part of VAEs and applying to the MNIST dataset.
\ No newline at end of file