Multi-Channel MHLF: LSTM-FCN using MACD-Histogram with Multi-Channel Input for Time Series Classification

Abstract

Time series classification is an important task for the identification of a person, weather, and motion, among others. In this study, the deep learning-based model is used for classification. In many research works, the deep learning-based time series classification has been reported to demonstrate a high performance. In particular, the LSTM-FCN model is a deep learning-based model, which shows the highest performance for time series classification. The proposed model is based on LSTM-FCN and its input consists of a multi-channel time series including the time series data and their MACD-histogram. In the experiments, the proposed model is evaluated using the defacto standard benchmark dataset, namely the UCR time series classification archive. The results show that the proposed model has higher performance than the existing models.