Overview of Time Series Classification Based on Symbolic Discretization for ECG Applications

This paper presents a comprehensive review of the Time Series (TS) data classification based on symbolic representation. In particular, we will focus on the classification of the Electro-Cardio Graphic (ECG) signal whatever the application it is used for. That is, we review the TS symbolic discretization in order to guide practitioners to make appropriate choice which meet their requirements in the ECG classification problem. We believe that scalability challenge in a real-time monitoring of cardiovascular patients, for example, require adapted representation that enhance dimensionality reduction before conducting classification process. Furthermore, an overview of the recent classification methods based on ordered symbolic attributes is detailed.