ABSTRACT
This chapter discusses methods that can be used to classify various cases, usually abnormal photoplethysmogram (PPG) and normal PPG signals. Different classifiers, especially neural networks, have been used extensively in different applications. The Bayes classifier is a simple “probabilistic” classifier, developed based on Bayes’ theorem. It is highly scalable, requiring a number of parameters linear in the number of variables (features/predictors) in a learning problem. This concept is crucial to understanding the core idea behind classification and pattern recognition. With the increasing availability of computing resources, deep learning has become a popular solution in computer vision, text recognition, speech translation, physiological data mining, and so on. If the initial feature set has a test set withheld, with development and training using the remaining set of features, the evaluation method is called ‘hold-out’ testing, which is the simplest form of validation.
