What is it about?
Generative ML models have been shown to work well for many data types such as images and text, but their effectiveness for time-series is unclear due to the lack of proper evaluation methods. We compare three generators for time-series with a case study of human activity recognition data. We also clearly explain to the reader a thorough and comprehensive evaluation process using a number of quantitative and qualitative evaluation metrics.
Featured Image
Photo by Mika Baumeister on Unsplash
Why is it important?
We provide this work for the community to encourage more people to work on time-series data generation to solve current limitations of data collection, cleaning and labeling.
Perspectives
I hope that this work will help researchers gain more understanding of time-series data generation and evaluation. Working on this project opened up new opportunities and insights for my PhD research.
Naif Alzahrani
Newcastle University
Read the Original
This page is a summary of: Experience: A Comparative Analysis of Multivariate Time-Series Generative Models: A Case Study on Human Activity Data, Journal of Data and Information Quality, August 2024, ACM (Association for Computing Machinery),
DOI: 10.1145/3688393.
You can read the full text:
Contributors
The following have contributed to this page
