DNCCLA: Discrete New Caledonian Crow Learning Algorithm for Solving Traveling Salesman Problem
The development of metaheuristic algorithms has led to the solution of various optimization problems. Bioinspired optimization algorithms like the New Caledonian crow learning algorithm (NCCLA) are primarily designed to address continuous problems. As most real-world problems are discrete, some oper...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2024-01-01
|
| Series: | Applied Computational Intelligence and Soft Computing |
| Online Access: | http://dx.doi.org/10.1155/acis/5324998 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The development of metaheuristic algorithms has led to the solution of various optimization problems. Bioinspired optimization algorithms like the New Caledonian crow learning algorithm (NCCLA) are primarily designed to address continuous problems. As most real-world problems are discrete, some operators have been proposed to convert continuous algorithms into discrete ones to address these problems. These operators include evolutionary operators such as crossover and mutation, transformation operators such as symmetry, swap, and shift, and K-opt algorithms such as 2-opt, 2-opt and a half, and 3-opt. Employing some of these operators usually accompanies changing the algorithm’s rules or the movement patterns of its search agents. However, mathematical operators such as modular arithmetic and set theory and random permutation provide an ability to keep the same algorithm’s agent proposed in its continuous version and K-opt algorithms usually balance the algorithm’s exploration and exploitation capabilities. Thus, this paper converts the NCCLA into a discrete version by utilizing a combination of those mathematical operators and the 3-opt algorithm. This combination allows the algorithm to maintain a balance between exploration and exploitation. The resulting algorithm, called the discrete New Caledonian crow learning algorithm (DNCCLA), is employed to solve the traveling salesman problem (TSP). In addition, the paper investigates the best combination of mathematical operators with K-opt algorithms or the symmetry operator. The performance results demonstrate that DNCCLA outperforms state-of-the-art discrete algorithms, exhibiting a good balance between exploration and exploitation. The algorithm successfully solves 20 TSP instances of varying scales, and it consistently achieved the top rank among the tested algorithms. |
|---|---|
| ISSN: | 1687-9732 |