Join our book community on Discord

https://packt.link/EarlyAccessCommunity

Up to this point, our recipes for training models have used datasets designed for easy class separability. These are great for illustration, but when we enter the real world, we’ll find that problems requiring machine learning (ML) solutions are rarely as cut-and-dry – we need additional tools in our toolchest for handling datasets where the boundaries between classes isn’t so well defined. In this chapter, we’ll explore methods that allow us to transform our dataset prior to training in a way that maps our original dataset onto a higher-dimensional representation allowing for easier class separability when using a straight line just won’t cut it (pun slightly intended). In this chapter, you will explore Support Vector Machines (SVMs) and kernel methods, focusing on theory, practical applications, and tuning techniques for high-dimensional data. Exercises include building SVM...