A hands-on tutorial in end-to-end ML systems development.

The purpose of this tutorial is to introduce students to how to manage end-to-end ML systems development. One key part is to manage machine learning experiments using MLFlow. This will consist of how to reproduce, track, and evaluate your experiments. Within an experiment, we will capture relationships among configurable parameters, machine learning code, the input data, output result, and performance metrics. In addition, we can also check the reproducibility of a machine learning algorithm.

To have an overview of the tutorial, the student can take a look at the tutorial link here:

In the first part, data preprocessing, students will have to run it by themself using prepared source code and following the detailed instruction from the git repository.

For training and serving ML models, it is recommended that students install the required software, libraries so that we can go smoothly with the tutorial. For instance, you could install Anaconda, python-pip, MLflow library if you want to run the experiments on your local machine. However, it is not mandatory, we can do the tutorial together on Google Cloud (you should install Visual Studio Code for remote code editing). The authentications details will be sent to you during the online session.

Last modified: Tuesday, 28 September 2021, 2:26 PM