Machine Learning Course at Sapienza University of Rome

Repository of the different challenges (midterm) during the Machine Learning course at Sapienza.

Currently, the code of the first challenge is present, which is worth 10% of the final score.

First Challenge

The first challenge as explained in the Jupyter notebook is to clean up a dataset, perform various dataset analyzes and show the comparison between the prediction power of the Decision Tree Classifier and the Multi-Layer Perceptron algorithm (MLP).
Grade: 30/30

Second Challenge

The second challenge is to develop a Convolutional Neural Network (CNN) to classify lego blocks. To develop it all I used the Keras and Tensorflow framework. The challenge also required to show the most significant pixels (features) through a heatmap.
Grade: 30/30 with honors.

Authors

• Andrea Bacciu - github