Sequence sort: A new non-dominated sorting algorithm for evolutionary multi-objective optimization

Area of Science:

• Computational Intelligence
• Optimization Algorithms
• Evolutionary Computation

Background:

• Non-dominated sorting is essential for multi-objective evolutionary algorithms (MOEAs).
• Existing algorithms face computational inefficiency and complexity, especially with more objectives.
• This limits their scalability and practical application in solving complex problems.

Purpose of the Study:

• To introduce Sequence Sort, a novel, efficient non-dominated sorting algorithm.
• To address the limitations of existing algorithms in terms of computational efficiency and complexity.
• To provide a reliable and scalable Pareto-based sorting method for MOEAs.

Main Methods:

• Developed Sequence Sort, incorporating presorting and solution marking strategies.
• Achieved a best-case computational complexity of O(N log N) for M objectives.
• Compared Sequence Sort against established algorithms like Fast Non-dominated Sorting, Deductive Sorting, HNDS, Corner Sorting, and MNDS.

Main Results:

• Sequence Sort demonstrates significant computational efficiency advantages over reference methods.
• Experimental results confirm sorting outcomes consistent with established algorithms.
• Statistical tests validate the significance of the performance improvements.

Conclusions:

• Sequence Sort offers a computationally efficient and reliable alternative for non-dominated sorting in MOEAs.
• The algorithm's performance is particularly notable as the number of objectives increases.
• It presents a promising advancement for tackling complex multi-objective optimization challenges.