The algorithm is then extended to use K-means clustering to seed the initial solution and the information pheromone is adjusted according to them.

对蚁群算法作了改进，思路是K-均值方法混合，利用K-均值方法的结果作为初值。