An improved adaptive memetic differential evolution optimization algorithms for data clustering problems

Area of Science:

• Computer Science
• Artificial Intelligence
• Optimization Algorithms

Background:

• Data clustering algorithm performance relies on balancing search exploration and exploitation.
• Existing algorithms show limitations in real-world dataset performance.
• Improved optimization strategies are needed for effective data clustering.

Purpose of the Study:

• To propose an adaptive memetic differential evolution optimisation algorithm (AMADE) for data clustering.
• To enhance the balance between exploration and exploitation in the search process.
• To improve the performance of data clustering on real-life datasets.

Main Methods:

• Developed an adaptive memetic algorithm (MA) incorporating a differential evolution (DE) mutation strategy.
• Hybridized the adaptive DE mutation operator with the MA to create AMADE.
• Evaluated AMADE on several real-life benchmark datasets.

Main Results:

• AMADE demonstrated faster convergence compared to standalone MA and DE.
• The proposed AMADE algorithm achieved superior performance over other clustering algorithms.
• Statistical analysis confirmed the effectiveness of AMADE on benchmark datasets.

Conclusions:

• Hybridizing memetic algorithms with adaptive differential evolution is effective for data clustering.
• AMADE successfully balances global exploration and local exploitation in optimization.
• The proposed approach offers a significant improvement for data clustering problems.