Applications of nature-inspired metaheuristic algorithms for tackling optimization problems across disciplines

Area of Science:

• Artificial Intelligence
• Statistics
• Bioinformatics
• Education Research
• Ecology
• Automotive Industry

Background:

• Nature-inspired metaheuristic algorithms are vital AI tools for complex optimization.
• These algorithms are increasingly adopted across various scientific disciplines.
• The competitive swarm optimizer with mutated agents (CSO-MA) is a novel, high-performing metaheuristic algorithm.

Purpose of the Study:

• To demonstrate the efficiency of the CSO-MA algorithm for diverse statistical optimization problems.
• To showcase the versatility of metaheuristic approaches in tackling real-world challenges.
• To compare metaheuristic performance against traditional statistical optimization methods.

Main Methods:

• Application of the competitive swarm optimizer with mutated agents (CSO-MA).
• Utilizing CSO-MA for parameter estimation in bioinformatics (single-cell generalized trend model).
• Employing CSO-MA for parameter estimation in educational models (Rasch model).
• Using CSO-MA for robust regression (Cox regression in Markov renewal models).
• Applying CSO-MA for missing data imputation (matrix completion in a two-compartment model).
• Leveraging CSO-MA for variable selection in ecological modeling.
• Optimizing experimental design using CSO-MA in the automotive industry (logistic model).

Main Results:

• CSO-MA demonstrated superior performance in various statistical optimization tasks.
• The algorithm successfully addressed complex problems in bioinformatics, education, ecology, and industry.
• Metaheuristics, including CSO-MA, showed potential to outperform standard statistical optimization algorithms.
• Effective imputation of missing data and optimal variable selection were achieved using CSO-MA.

Conclusions:

• Nature-inspired metaheuristic algorithms, exemplified by CSO-MA, are highly effective for a wide range of statistical optimization problems.
• CSO-MA offers a flexible and powerful tool for researchers and practitioners across multiple disciplines.
• The study highlights the significant potential of metaheuristics to enhance statistical modeling and problem-solving.