Implement in PyTorch an autoencoder to learn data representation. Different transfer learning techniques are then compared to classify a labelled images.
This assignment is developed in Python 3 for the Advanced Topics in Machine Learning course (given by Prof. Dr. Paolo Favaro) of the Master in Computer Science in the university of Bern, Switzerland.
A large unlabelled training set of images as well as a small labelled training set were given. The objective was to learn data representation with some unsupervised feature learning method of our choice and then compare different variations of transfer learning techniques.
This assignment was an individual work.
Spring 2020.