A Modified Ant Colony Algorithm for Traveling Salesman Problem

Ant colony algorithms such as Ant Colony Optimization (ACO) havebeen effectively applied to solve the Traveling Salesman problem (TSP). However,traditional ACO algorithm has some issues such as long iterative length and prone tolocal convergence. To this end, we propose we embed ACO into Cultural Algorithm(CA) framework by leveraging the dual inheritance mechanism. Best solutions areevolved in both population space and belief space, and the communication betweenthem is achieved by accept and influence operations. Besides, we employ multiplepopulation spaces for parallel execution. Experiments show that the performance ofour proposed algorithm is greatly improved.