DeepLearning

Projects from the Deep Learning Course of EPFL (EE-559)

It contains two directories, each one corresponding to 1 project:

Proj1: Project 1 - Classification, weight sharing, auxiliary losses

Contains the first project.

Goal

Given two digits, we train neural networks to recognize if the first one is smaller or equal to the second. Here various methods are used to study the effect of weight-sharing and auxiliary losses.

Proj2: Project 2 - Mini deep-learning framework

Contains the second project.

Goal

In this project, we implemented a mini-deep learning framework using only basic tensor operations and the python's math library.