Competitive Swarm Optimizer with Mutated Agents for Finding Optimal Designs for Nonlinear Regression Models with...

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Metaheuristic Computing

Background:

• Existing metaheuristic algorithms, including the Competitive Swarm Optimizer (CSO), face challenges in maintaining swarm diversity and balancing exploration-exploitation.
• Effective optimization is crucial for solving complex problems, particularly in high-dimensional statistical modeling and experimental design.

Purpose of the Study:

• To introduce a novel enhancement to the Competitive Swarm Optimizer (CSO) called CSO with Mutated Agents (CSO-MA).
• To improve swarm diversity and space exploration capabilities of the CSO algorithm.
• To demonstrate the effectiveness of CSO-MA in solving high-dimensional optimal design problems and compare its performance against other state-of-the-art algorithms.

Main Methods:

• The proposed CSO-MA algorithm mutates 'loser' particles within the swarm to enhance diversity and exploration.
• A selection mechanism is implemented to ensure that exploration in promising areas is not hindered.
• Performance is evaluated through simulations against CSO, other swarm-based algorithms, and the Cuckoo search algorithm.

Main Results:

• CSO-MA demonstrates a superior exploration-exploitation balance compared to the original CSO.
• CSO-MA generally outperforms CSO and other swarm-based algorithms, as well as the Cuckoo search algorithm.
• The algorithm successfully addressed a high-dimensional optimal design problem where other swarm algorithms failed.

Conclusions:

• CSO-MA is a robust and effective optimization tool that enhances swarm diversity and exploration.
• The algorithm offers a general-purpose solution applicable to various optimal design problems, including those for nonlinear models.
• CSO-MA provides a competitive alternative to existing metaheuristic algorithms without significantly increasing computational cost.