Supervised and unsupervised ML

The goal of ML is to develop and deploy computational algorithms that can automatically learn and improve from experience without human interference to perform a particular task. But how does it work? It does so by first “learning” knowledge from experience from the input data and using that knowledge to make predictions on unseen data. As such, the crux of ML is the learning problem in which machines learn from real-world data, improve from experience, extract patterns, construct models, and predict the outcomes of unseen data.

Depending on the type of data and the tasks to perform, ML algorithms can be broadly divided into supervised, semi-supervised, and unsupervised methods. Supervised methods learn patterns from examples with labels (for example, “diseased” or “not diseased”) and are then used to predict future events or labels from unseen data (Figure 3.1). Unsupervised methods, in contrast, don’...