Bi-Directional Long Short Term Memory with Dynamic Time Wrapping for Time Series Classifications

Time series data includes both global and regional patterns. Most currently used nature networks concentrate on local features rather than their connections. The latter is likewise crucial but more problematic to investigate due to the difficulty in employing feature networks to gather enough representations. This research focus to design a new model with the help of “Dynamic Time Warping (DTW)” and Bidirectional Long Short-Term Memory (Bi-LSTM). This model encompasses processes such as gathering data, extracting features, and performing classification. Here, DTW is used for the feature extraction to get the significant features. Then, a DTW -based extracted feature is fed to the classification stage, the classification results are obtained through the utilization of “Bidirectional Long Short-Term Memory” Further, the experimental results have verified the superior performance of the designed proposed model over other existing time series classification approaches.