Data Clustering Using Moth-Flame Optimization Algorithm

Area of Science:

• Computer Science
• Artificial Intelligence
• Optimization Algorithms

Background:

• K-means clustering performance is sensitive to initial cluster centers and prone to local optima.
• Metaheuristic algorithms offer robust solutions for complex optimization problems.
• The Moth Flame Optimizer (MFO) is a recent metaheuristic demonstrating strong performance in various applications.

Purpose of the Study:

• To propose a novel heuristic clustering approach leveraging the Moth Flame Optimizer (MFO).
• To address the limitations of traditional k-means algorithms in data clustering.
• To evaluate the effectiveness and competitiveness of the MFO-based clustering method.

Main Methods:

• A new heuristic clustering algorithm is developed using the principles of the Moth Flame Optimizer (MFO).
• The proposed MFO-based clustering approach is tested on Shape and UCI benchmark datasets.
• Experimental validation involves comparing the MFO algorithm against five state-of-the-art clustering algorithms.

Main Results:

• The MFO-based clustering approach achieved superior mean performance on 10 out of 12 datasets.
• The algorithm showed comparable performance on the remaining two datasets.
• Experimental results confirm the efficacy and robustness of the proposed MFO clustering method.

Conclusions:

• The proposed Moth Flame Optimizer (MFO) heuristic effectively solves data clustering problems.
• This novel approach offers a competitive alternative to existing state-of-the-art clustering algorithms.
• The MFO-based method shows significant potential for improving clustering accuracy and efficiency.