Metaheuristic clustering
- 251 pages
- 9 hours of reading
Cluster analysis organizes an unlabeled collection of objects into distinct groups based on similarity. This process has been explored through various domains, including graph theory, multivariate analysis, neural networks, and fuzzy set theory. While clustering is often labeled as an unsupervised learning method, many traditional algorithms require prior specification of the number of clusters, which limits its unsupervised nature. Modern data mining tools necessitate rapid and fully automatic clustering of large datasets with minimal user intervention to predict future trends and behaviors for informed decision-making. This volume presents clustering as an optimization problem, aiming to find the best partitioning of a dataset by minimizing or maximizing one or more objective functions. It showcases real-world applications and demonstrates the effectiveness of several metaheuristics, particularly the Differential Evolution algorithm, in both single and multi-objective clustering scenarios where the number of clusters is not predetermined. Comprising seven chapters, the book starts with fundamental definitions and concludes with significant research challenges. It offers valuable insights for academics, scientists, and engineers involved in optimization techniques and data mining research and applications.
