Time series classification using local distance-based features in multi-modal fusion networks

作者：

Highlights：

• Proposes a method of using the local distances between time series as features.

• Uses multi-modal CNNs to combine the raw coordinates and local distance features.

• Studies the effects of that prototype number and selection have on accuracy.

• Validates the proposed method on Unipen, UCI, and UCR time series datasets.

摘要

•Proposes a method of using the local distances between time series as features.•Uses multi-modal CNNs to combine the raw coordinates and local distance features.•Studies the effects of that prototype number and selection have on accuracy.•Validates the proposed method on Unipen, UCI, and UCR time series datasets.

论文关键词：Convolutional Neural Network,Time series classification,Dynamic time warping,Distance features

论文评审过程：Received 14 January 2019, Revised 10 June 2019, Accepted 26 August 2019, Available online 26 August 2019, Version of Record 30 August 2019.

论文官网地址：https://doi.org/10.1016/j.patcog.2019.107024