Time Series Classification Using Convolutional Neural Network On Imbalanced Datasets

TL;DR

This paper investigates the challenge of class imbalance in time series classification and evaluates various sampling and algorithmic methods to improve performance, achieving high F scores even with skewed data.

Contribution

It introduces and compares different approaches to handle class imbalance in TSC, demonstrating significant performance improvements over traditional methods.

Findings

Sampling and algorithmic methods improve TSC accuracy on imbalanced datasets.

F score reaches up to 97.6% on simulated datasets despite high imbalance.

Methods are effective across different types of time series data.

Abstract

Time Series Classification (TSC) has drawn a lot of attention in literature because of its broad range of applications for different domains, such as medical data mining, weather forecasting. Although TSC algorithms are designed for balanced datasets, most real-life time series datasets are imbalanced. The Skewed distribution is a problem for time series classification both in distance-based and feature-based algorithms under the condition of poor class separability. To address the imbalance problem, both sampling-based and algorithmic approaches are used in this paper. Different methods significantly improve time series classification's performance on imbalanced datasets. Despite having a high imbalance ratio, the result showed that F score could be as high as 97.6% for the simulated TwoPatterns Dataset.