Exploration of Multi-objective Soft Subspace Clustering Based on Data Reliability
Lina Mao, Linyan Tang · 2024
With the advent of the era of information and data, a large amount of data has been accumulated in various fields. How to use these data effectively and tap its potential value has become an urgent problem. Cluster analysis is an important method in data mining, its purpose is to group similar objects into one class, so that objects in the same class are as similar as possible, and objects in different classes are as different as possible. However, traditional clustering methods are often based on full space, which may lead to unsatisfactory results in the face of complex data sets, especially in high-dimensional space. Therefore, subspace clustering has become a research hotspot in recent years. The purpose of subspace clustering is to find the local structure of the data, so as to obtain more accurate clustering results. In addition, the reliability of data is also a factor that cannot be ignored in cluster analysis. The quality of data is often uneven due to possible errors and noise during data collection, storage and processing. Using this data directly for clustering can result in incorrect clustering results. Therefore, how to effectively consider data reliability and integrate it into subspace clustering algorithm is a problem worth studying. In this paper, a filtered multi-objective soft subspace clustering algorithm framework R-MOSSC based on data reliability is proposed, and an algorithm example AMGA2SC is given.