Deep Learning Projects

Emilio Fernández, Alejandro Bonell, Berk Mandiracioglu

This repository contains the two required projects for EPFL Deep Learning course:

• In the project ConvolutionalNNs, we present and explore different deep neural networks architectures for distinguishing two different numerical digits from the MNIST database. We highlighted the advantages of using convolutional layers, weight sharing and auxiliary losses.

• In the project NeuralFramework, we developed a custom Pytorch deep learning framework for training diverse neural networks using tensors. We proved the correct functioning of our framework by developing a simple machine learning model