Comparison of the Learning Mechanism between Genetic Algorithm and Ant Colony Optimization Algorithm

the traditional optimization approaches require the problem-specific knowledge, so they fail to solve the black-box optimization problem. The adaptive sampling search algorithms can effectively solve these problems by exploiting the information gathered from samples taken from the search space. This paper firstly investigates the relationship between learning and optimization, and points out the essentiality of learning in Black-box Optimization. Then the searching mechanisms of genetic algorithm (GA) and ant colony optimization (ACO) algorithm are analysized in the perspective of machine learning, and conclude that GA belongs to instance-based learning while ACO belongs to model-based learning. Based on TSP, experimental analyses are performed to compare the learning ability between GA and ACO.