Welcome back to this series looking into the dark arts of machine learning, last time we introduced the concept of cost functions and how they can be used to optimize machine learning algorithms. Here we’re going to explore this powerful technique in more detail. Enjoy!
So before we figure out what a PCF is, it is worth defining what a cost function is. Generally speaking, supervised learning can be typically broken down into 4 main components.
3. Cost function
4. Optimization goal
In this example we are actually looking at a very simple linear regression model…
How can we learn in an unsupervised world?
In our first post in this series, we introduced the challenges and opportunities that unsupervised machine learning hold. In this post, we’ll start with a quick refresher on evaluation metrics, and then explore how we can build models and learn in this space.
When learning from a dataset in a supervised setting, we want to compare our predictions to reality. And similar to when we decide on a new washing machine & consider the energy efficiency rating, we want a simplified representation of performance, from which we can make decisions from.
Welcome to this series into the Dark Arts of Unsupervised Machine Learning. Over the coming weeks, we’ll post the remaining lessons, so keep an eye out! We hope you enjoy!
So before we start, it’s worth explaining a bit about what unsupervised learning really is and why we might want to use it. …