A New Two-Stage Algorithm for Solving Optimization Problems

Area of Science:

• Computational Intelligence
• Machine Learning Algorithms
• Optimization Techniques

Background:

• Optimization is crucial for finding maximum or minimum values of objective functions.
• Existing optimization algorithms often draw inspiration from natural phenomena.
• Algorithm-based optimization is fundamental to machine learning and artificial intelligence.

Purpose of the Study:

• To introduce a new optimization algorithm named two-stage optimization (TSO).
• To mathematically model and describe the stages of the TSO algorithm.
• To evaluate the performance of TSO against established optimization methods.

Main Methods:

• The TSO algorithm employs a two-step update process for population members in each iteration.
• A subset of high-performing members is selected.
• Two randomly chosen members from this subset are used sequentially to update each member's position.

Main Results:

• TSO was evaluated on twenty-three standard objective functions.
• Its performance was compared against eight other algorithms, including genetic, particle swarm, and grey wolf optimization.
• Numerical results indicate TSO's superiority and competitiveness.

Conclusions:

• The proposed two-stage optimization (TSO) algorithm demonstrates significant advantages.
• TSO offers a competitive and effective approach for solving various optimization problems.
• This new algorithm shows promise for advancing machine learning and AI applications.