Project Zoidberg

The goal of this project is to, given some X-ray images, use machine learning to help doctors detecting pneumonia. We have access to 3 datasets available here.

We must:

• use a train-validation-test procedure
• use a cross validation procedure,
• compare our results with a simple train test split,
• use one of the datasets to tune our algorithms.

We deliver:

• a technical documents: a Jupyter notebook-like file, containing code and text, possibly graphics and an html-file to prove our results without rerunning the code notebook
• a synthesis document: a pdf file to sum up our results and figures pdf

We choose to use the following Machine Learning (ML) algorithm:

• k-Nearest Neighbors (KNN)
• Support Vector Machine (SVM)
• Naive Bayes (NB)
• Convolutionnal Neural Network (CNN with keras (tensorflow))