分类
计算机科学
情报检索
系列(地层学)
数据科学
相似性(几何)
系统回顾
数据挖掘
机器学习
人工智能
主题(文档)
万维网
梅德林
古生物学
政治学
法学
图像(数学)
生物
作者
Wildan Mahmud,Ahmad Zainul Fanani,Heru Agus Santoso,Fikri Budiman
标识
DOI:10.1109/isemantic59612.2023.10295319
摘要
In order to spot trends in the methodologies and procedures employed, this systematic literature review will look at works on time series categorization. Six research questions are used as a guide to perform a systematic literature evaluation based on the PICOC criteria. In order to find articles that fit the required criteria, a search is done through trustworthy article database providers. One article from the collection was picked as the main one, which will be used to find further articles with information that is similar. A search based on the main article turned up 115 linked articles. According to the systematic literature review's findings, there were 28 articles on time series categorization in 2020, but only 19 by mid-May 2023. This suggests a high rate of publishing, illustrating the academics' interest in the study of time series categorization. Time series classification research articles are regularly published in the Springer Publisher magazine Data Mining and Knowledge Discovery, followed by IEEE Access. Public datasets that are utilized as experimental datasets often at rates of 35% and 10%, respectively, include the UCR archive and the UAE archive. This shows that experiments regularly use UCR datasets as references. Over the last five years, deep learning has been the subject of 30% of research papers, while random convolutional has contributed 14% of publications and is a promising research trend. Time series can be categorized using methods based on similarity, interval, shapelet, dictionary, deep learning, random convolution, and combination.
科研通智能强力驱动
Strongly Powered by AbleSci AI