Abstract

Time series classification (TSC) is a significant problem in data mining with several applications in different domains. Mining different distinguishing features is the primary method. One promising method is algorithms based on the morphological structure of time series, which are interpretable and accurate. However, existing structural feature-based algorithms, such as time series forest (TSF) and shapelet traverse, all features through many random combinations, which means that a lot of training time and computing resources are required to filter meaningless features, important distinguishing information will be ignored. To overcome this problem, in this paper, we propose a perceptual features-based framework for TSC. We are inspired by how humans observe time series and realize that there are usually only a few essential points that need to be remembered for a time series. Although the complex time series has a lot of details, a small number of data points is enough to describe the shape of the entire sample. First, we use the improved perceptually important points (PIPs) to extract key points and use them as the basis for time series segmentation to obtain a combination of interval-level and point-level features. Secondly, we propose a framework to explore the effects of perceptual structural features combined with decision trees (DT), random forests (RF), and gradient boosting decision trees (GBDT) on TSC. The experimental results on the UCR datasets show that our work has achieved leading accuracy, which is instructive for follow-up research.

Highlights

  • In the information age, massive amounts of data have been generated over time

  • This paper proposes a classification framework based on perceptual features, which can extract support points of morphological structure from the original time series and further obtain interval-level and point-level features for classifiers such as decision trees

  • The following five classification algorithms were selected for comparison, including the word extraction for time series classification (WEASEL), bag of symbolic-fourier approximation symbols (BOSS), time series forest (TSF), random interval spectral ensemble (RISE), and canonical time-series characteristics (Catch22)

Read more

Summary

Introduction

Massive amounts of data have been generated over time. These data are closely related to many studies. The Bag-of-SFA Symbols (BOSS) method based on the bag-of-words model was proposed [14] This method records high-frequency symbol features and uses them to distinguish different types of time series samples. They proposed ROCKET, a kernel-based time series classification method This is a new direction for TSC, which can both reduce computational complexity and improve accuracy. This paper proposes a classification framework based on perceptual features, which can extract support points of morphological structure from the original time series and further obtain interval-level and point-level features for classifiers such as decision trees. The data points extracted by GRM-PIPs can divide the time series into sub-sequences similar to shapelets. PIPs uses a concise idea to extract important points in the morphological structure of time series.

Decision Tree and Ensemble Methods
Feature Extraction
Classifer and the PFC Framework
The Verification of Two-Category
The Hybrid Verification
Findings
Discussion on the Number of PIPs
Conclusions
Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call