I was reading the great Tom Mitchel's book on machine learning and I came across one of its examples which were about facial recognition using artificial intelligence. The book has provided us with C++ code, yet I thought it would be useful to write the code in Python. (I searched but could not find a Python code) So I wrote the code, and I have assumed that python will do nothing special for backpropagation learning and I wrote the code from scratch. In this manner, the code is longer and maybe harder to read. Yet, it shows a beginner how exactly the backpropagation works. Also, you should try different hyper-parameters in the code ( learning rate and epochs) to achieve desirable results.

Using the MATLAB deep learning toolbox to successfully code a convolutional neural network from scratch and Python to extract and clean the chest X-ray dataset which is needed to train the neural network. There are many subsections that make up a successful neural network, such as supervised learning, image processing, transfer learning, backpropagation, parallel computing, data cleaning and extraction which are all important aspects to understand before creating a successful neural network that could extract features from 8 different thorax classes. This project was completed in the space of 5 months, with no prior knowledge of deep learning using my dedicated prebuilt PC.