What is it about?
Unsupervised machine learning (ML) methods are incorporated in this work to find correlations in a dataset and investigate hidden relations between data points. Widely used clustering algorithms are incorporated, such as the k-means, k-medoids, and DBSCAN, to create data frames of common characteristics. The symbolic regression (SR) ML method is applied, next, on each cluster and provides analytical expressions that describe the fluid’s behavior under various temperature and density conditions, suggesting an alternative computatinal pathway.
Featured Image
Photo by Wim van 't Einde on Unsplash
Why is it important?
In fluid mechanics and materials science, in general, in which experimental techniques are hard and expensive to perform and simulations are computationally demanding, ML techniques are also incorporated. The present research is focused on reassessing well established equations and methodologies, led by unsupervised machine learning techniques and by exploiting the power of symbolic regression, a genetic programming technique that extracts symbolic expressions from an infinite phase space based only on data, without prior knowledge of the system. These equations can be used for the accurate description as well as the discovery of the underlying mechanisms.
Perspectives
We believe that the two-fold clustering/symbolic regression method presented here has a prominent potential for material investigation and research, and could be incorporated for extracting and predicting properties as an embedded method in classical computational frameworks.
Dr. Filippos Sofos
University of Thessaly
Read the Original
This page is a summary of: Calculating material properties with purely data-driven methods, September 2022, ACM (Association for Computing Machinery),
DOI: 10.1145/3549737.3549802.
You can read the full text:
Resources
Contributors
The following have contributed to this page
