This repository contains the work on my Individual Project, entitled "Experimental Study on the Properties of Disentangled Autoencoders", implemented during my final year (Part II) of the Computer Science Tripos at the University of Cambridge. I would like to express my sincere gratitude to my supervisors, Petar Veličković and Dr Pietro Liò, for their invaluable advice and endless support. The results from this dissertation were extended and led to a paper.
Peychev, M., Veličković, P. and Liò, P. (2017) Quantifying the Effects of Enforcing Disentanglement on Variational Autoencoders. Workshop on Learning Disentangled Features at the 31st Annual Conference on Neural Information Processing Systems (NeurIPS 2017). [arXiv]