I. Introduction
In the last decades the interest in mining massive and potentially unbounded datasets which arrive at rapid rates, namely data streams, has grown substantially. Mining data streams is of the utmost importance since many available data generators produce enormous amounts of data over time. Examples of data streams include sensor networks, wearable sensors, computer network traffic sniffers and video surveillance, to name a few. Aiming at extracting useful knowledge from these massive amounts of data, a variety of inductive learning techniques were developed and achieved concrete results in both supervised [1], [2] and unsupervised [3]–[6] tasks.