Welcome to my introductory course on Testing Machine Learning Models.
My name is Carlos Kidman and I wanted to talk about machine learning from the perspective of a tester.
Machine learning and artificial intelligence are becoming more prevalent as companies look to leverage their data to solve problems that traditional programming cannot.
Data science and machine learning teams are looking for testing professionals to help them with their data, their pipelines, models, services, AI ethics, and more as they build these solutions.
However, machine learning is a different paradigm than traditional programming and some of the ways you test will be a little bit different as well.
In this course, we'll be looking at real machine learning models and how we can test them.
We'll be using Python as well since it's the most popular language for ML and AI.
We'll start by demystifying machine learning and talking about the difference between ML and traditional programming.
Then we'll train and build our first ML model in a tool called Google Colab. With that context, we'll be able to explore just where testers fit in the world of machine learning.
We find out that testers fit quite nicely, so we'll be looking at how we can perform adversarial attacks to fool these models.
Then we'll look at different behavioral testing techniques that we can use as well.
That will naturally lead us into how we can test these ML systems to make sure that they are fair and responsible instead of causing harm, like unfair bias.
And lastly, we'll wrap things up by looking at some models that have been deployed to production.
Machine learning is a deep and vast field, so we won't be covering every model type and architecture, but by the end, you will have learned ideas, concepts, techniques, tools, strategies, all of that to effectively test ML models in the real world.
So, let's get started.