InceptionTime vs. Wavelet -- A comparison for time series classification

TL;DR

Neural networks were used to classify infrasound data, using a custom implementation of the InceptionTime network and a ResNet implementation to generate 2D images of the wavelet transformation of the signals.

Abstract

Neural networks were used to classify infrasound data. Two different approaches were compared. One based on the direct classification of time series data, using a custom implementation of the InceptionTime network. For the other approach, we generated 2D images of the wavelet transformation of the signals, which were subsequently classified using a ResNet implementation. Choosing appropriate hyperparameter settings, both achieve a classification accuracy of above 90 %, with the direct approach reaching 95.2 %.

Figures (1)

• Figure 1: Raw signals and corresponding wavelet transformation of two different time series signals out of category 0 and 5.