Multi-Strategy Enhanced Parrot Optimizer: Global Optimization and Feature Selection.

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Machine Learning

Background:

• Optimization algorithms are crucial for solving complex problems in global optimization and feature selection (FS).
• Existing algorithms may face limitations in adaptability and information exchange efficiency.

Purpose of the Study:

• To introduce the Enhanced Crisscross Parrot Optimizer (ECPO), an advanced optimization algorithm.
• To evaluate the ECPO's effectiveness in global optimization and feature selection tasks.

Main Methods:

• Developed the Enhanced Crisscross Parrot Optimizer (ECPO) with strategy selection and crisscross mechanisms.
• Evaluated ECPO on CEC2017 benchmark functions against ten conventional optimization algorithms.
• Applied a binary version of ECPO to feature selection problems on ten real-world datasets.

Main Results:

• ECPO demonstrated superior performance compared to ten other optimization algorithms on CEC2017 benchmark functions.
• The binary ECPO achieved competitive error rates with reduced feature subsets on real-world feature selection tasks.
• ECPO showed enhanced exploration capabilities due to improved information exchange.

Conclusions:

• The Enhanced Crisscross Parrot Optimizer (ECPO) is a highly effective algorithm for global optimization.
• ECPO shows significant promise for feature selection, offering competitive performance with fewer features.
• The proposed algorithm offers a robust and adaptable solution for complex optimization challenges.