Auto-Grouping Sedimentation Using Unsupervised Based Clustering Techniques

A Machine Learning (ML) algorithm plays an important role in the prediction of inaccuracies in several fields, such
as medicine, computer science, along underwater particle sedimentation. Hence, in this research work, the authors implemented various clustering methods for grouping the sediment particles such as mud, sand50, gravel50, rock 10 cm, rock 50cm, surface carbon, and nitrogen in the underwater sea automatically. This research focuses on the application of unsupervised machine learning, specifically clustering techniques, to automate the grouping of underwater sediment particles. The research highlights
the utilization of K-means Clustering and BIRCH Clustering, introducing a novel contribution in the form of a Hybrid Clustering approach that integrates the benefits of both methods. This hybridization is designed to refine and enhance clustering results, presenting a promising solution for the automation of sediment analysis in underwater environments. To predict the performance of various unsupervised machine learning-based clustering algorithms, metrics like Calinski Harabasz, Silhouette Score,
Mathew’s Correlation Score, Davies Bouldin, Hamming loss, and Cohen Kappa score with n=7 are evaluated in underwater
sediment particles grouping. Among several clustering techniques, the proposed hybrid approach outperforms in clustering of sediment articles based on the Silhouette score.