Swarm intelligence (SI) Colony of Ants or other social insects are: Social intelligence: simple individual behaviour but joint effect can be intelligence Decentralized: no central control of the individuals of the colony Self-organized: individual adapts to environment and other members of colony Robust: task is completed even if some individuals fail Main working principles of ants: Leaving pheromen tracks to paths between nest and food source Joint efforts to carry loads Solving Travelling Salesman problem by ants: “Sending” ants from source to explore different randomly chosen routes Short links are chosen more often than long ones in the route After receiving candidate route, good tracks are marked by “pheromen” During next iterations: tracks with high pheromen are chosen more often. Example (page on other side): 1. Four randomly chosen routes with the length marked inside. 2. Let’s add pheromen in the following way: Subtract cost of the links in the best solution (-1) and increase the ones in the worst solution (-1) 3. Repeat the procedure by taking new set of random routes. But remember, smaller links are be taken with higher probability! Costs are in respect to the original graph. 4. Let’s add pheromen again. Resulting graph is shown in the last. 5. The most likely route will converge towards the optimum (although optimality cannot be guaranteed). “Traveling sales-ants” for solving TSP B D C B A B 23 21 F C D E G B F A C B 26 24 D E H G C B F A C D E H G C B F A D E H G H C B C F 3 3 H B 0 C E 0 G 0 H B 2 C D 5 1 5 3 5 1 F +1 +1 0 -1 D -1 0 +1 -1 E 4 21 F A 22 D E H F A 23 22 D E G H F A D E G H F A D E G H 4 2 2 3 B 0 C B 3 D 5 0 A 6 E 4 1 H F 1 0 5 E C 4 0 -1 -1 G 0 -1 D +1 0 0 2 G 2 H 5 G 1 H +1 2 0 G +1 A 4 2 -1 A E G +1 A 3 2 5 B 2 3 C 4 4 A 2 2 F F