Many biomedical signal processing applications involve distinguishing between signals generated under different conditions. We might want to differentiate between the EEG signals from people when they are having a seizure and the signals when they are not having a seizure. Or, perhaps, we want to differentiate between EKG signals indicating tachycardia from normal EKG signals. Given a set of measurements, we want to select the process that generated those measurements from a set of candidate processes. The selection is based on the features of the processes that distinguish them from one another. If these features are measurable, we can construct a classifier that categorizes observations into distinct classes. We will talk about classifiers and their properties in a later chapter. For now, let’s focus on the idea of features.