Time Series Representation Techniques on High Dimensional Challenges and Computational Cost Increase: A Survey

الملخص

In recent times, time series data has gained a lot of attention and the aspects that are included in it are increasing, in addition to the growth of data flow in various fields such as transportation, weather, industry, finance, medicine, and entertainment, etc. Therefore, it became necessary to work on reducing the dimensions in order to overcome the limitations that occupy the available memory and to analyze this large number of information and data with good efficiency. Because of the high dimensionality and the high correlation of the large amounts of noise it generates, the researchers focused on proposing representations for time series and proposed many new representation schemes for this type of information and data for the purpose of addressing the challenges caused by high dimensionality. In addition to mining time series data that search for similarities and learn all data tasks, These tasks require a lot of time for the purpose of computation, which can be solved by reducing the dimensions of the data. Many researchers have published papers that suggest different measures of similarity based on representation for the purpose of solving the problem of high dimensions, so this paper will be an overview of the field of representation in the extraction of time series data and some recent trends and methods about it.