Hands On: AI-Driven Parkinsons Diagnosis

A workshop as introduction to Deep Learning and speach analysis.

During the workshops the participants will learn:

• The basic of neural networks, e.g. activation functions, loss functions
• The more advanced Recourent Neural Networks and LSTMs
• How to apply the gained knowledge to speach datasets

Agenda

We recommend the following agend.

1. System Setup
   You will need to bring your own laptop. There are two ways to run code during the workshop.
   a. Locally on your machine. There for you will need Python 3.6, Keras, and Tensorflow installed.
   b. Azure Notebooks You can run the code online over Azure Notebooks. https://notebooks.azure.com/
   
2. Introduction to machine learning
   This section will provide you with the basic iddeas of machine learning.
3. Introduction to deep learning
   All basics you need to know to understand what we are doing in the applied sections. We will cover the basic idea behind neural network including the most important elements, such as activations, loss, optimization, and the evaluation of your results.
4. Application to the spoken digits dataset
   
5. Advanced topic: RNN, LSTM
   Usually we can get better results using more advanced architecutures then just a simple fully connected network. In this section we will teach you the basic idea of RNNs and LSTMs.
6. Application to the Parkinsons Dataset
   Finally you will build a neural network your self which will classify whether a patient has Parkinsons.

For participants

If you will participate in the workshop, please checkout our setup instructions.

For instructors

If you want to hold this workshop by your self checkout our instruction on how to prepare this workshop.

Feedback

In case you have any questions regarding the workshop or the code it self, feel free to create a new issue

Contributors

• Henry van der Vegte [Github]
• Max-Philipp Schrader [Github]