Fair Clustering Ensemble With Equal Cluster Capacity.

Area of Science:

• Data Mining
• Machine Learning
• Artificial Intelligence

Background:

• Clustering ensemble methods are widely used but lack fairness considerations.
• Fairness is crucial in real-world applications, particularly those involving human data.

Purpose of the Study:

• To propose a novel fair clustering ensemble method.
• To address the limitations of existing methods regarding fairness and cluster imbalance.

Main Methods:

• Developed a new definition of fairness incorporating cluster capacity equality.
• Designed a simple yet effective regularized term to achieve fairness and capacity equality.
• Integrated the regularized term into a clustering ensemble framework.

Main Results:

• The proposed method achieves comparable or superior clustering performance.
• The method significantly improves fairness and cluster capacity equality compared to state-of-the-art methods.
• Experimental results demonstrate the effectiveness and superiority of the new approach.

Conclusions:

• The novel fair clustering ensemble method effectively balances fairness and clustering performance.
• The proposed definition of fairness and regularization term offer a robust solution for equitable clustering.
• This work advances the field of fair machine learning in clustering applications.