The Time Series Data Classification Method Based on Deep Spatial Transformer Network

In this paper, we propose a deep spatial transformer network (DSTN) to classify the time series data. This DSTN model can avoid the distortion thatmay be caused when the time series data is transformed to 2D-data, as it has an adaptive feature extractionmechanism profit and can carry out affine transformation on 2D-data. It can also avoid the influence of this distortion on the subsequent convolutional neural networks. Thus, Experimental results show that this framework can effectively classify and predict complex multivariable time series data.