Skills II: Developing, Software Stacks and Workflows

This module is easily the most technical of the course. We hope that at whatever level you are participating that you will still pull something from the information provided. So, dig into the Research Springboard or simply “catch as catch can.” Once again there are two videos within this module. The first is Skills Part 2 which picks up where the last Skills video encounters an issue. We look at why software like Docker and containerization has become very big not only in ML but in the internet as a whole. We explore development environments and introduce the concept of dependencies and how they can benefit but also complicate the development process. The second video looks at an exciting horizon in machine learning: federated or distributed learning. Distributed learning can remove barriers to ML by removing the need for access to expensive hardware. After this module you should be able to:

  • Explain what issues containerization solves in regard to training ML models
  • Define a software dependency and why they exist in open source software
  • Discern the steps required to train an open source machine learning model
  • Identify skills and tools required for continued learning
  • Define distributed or federated learning what issues it could impact