A series of notebooks implementing a suite of generative models of images.
These are meant to serve as:
- A refresher exercise for myself on unsupervised learning models and Tensorflow (especially TF2)
- A set of reference implementations of some basic ML components that I need for my own research.
Libraries and tools used:
- All of these are Google Colab Notebooks: https://colab.research.google.com/notebooks/intro.ipynb
- All of them are written in Tensorflow (mostly TF2, but some are in TF1)
- Some use the Sonnet library of abstractions for ML research https://sonnet.readthedocs.io/en/latest/, and some use the Keras library for deep learning https://www.tensorflow.org/guide/keras
To date I've implemented the following generative models:
Variational Autoencoder
- Original paper: https://arxiv.org/pdf/1312.6114.pdf
- Great explanation: https://jaan.io/what-is-variational-autoencoder-vae-tutorial/