A Literature Survey Of Early Time Series Classification And Deep Learning

Abstract

This paper provides an overview of current literature on time series classification approaches, in particular of early time<br> series classification. A very common and effective time series classification approach is the 1-Nearest Neighbor classier, with different distance measures such as the Euclidean or dynamic time warping distances. This paper starts by reviewing these<br> baseline methods. More recently, with the gain in popularity in the application of deep neural networks to the eld of computer vision, research has focused on developing deep learning architectures for time series classification as well. The literature in the field of deep learning for time series classification has shown promising results. Early time series classication aims to classify a time series with as few temporal observations as possible, while keeping the loss of classification accuracy at a minimum. Prominent early classification frameworks reviewed by this paper include, but are not limited to, ECTS, RelClass and ECDIRE. These works have shown that early time series classification may be feasible and performant, but they also show room for improvement.