Abstract
Ant Colony Optimization (ACO) has been used successfully to solve hard combinatorial optimization problems. This metaheuristic method is inspired by the foraging behavior of ant colonies, which manage to establish the shortest routes to feeding sources and back. On this work we use estimation of start nodes with respect to the quality of the solution. Various start strategies are offered. Sensitivity analysis of the algorithm behavior according strategy parameters is made. Our ideas is applied on Multiple Knapsack Problem (MKP) like a representative of the subset problems.