What is an Ant colony optimization algorithm?

What can we learn from the "ant colony optimization" algorithm to improve our problem solving skills?

• http://en.m.wikipedia.org/wiki/Ant_colony_optimization_algorithms

• Answer:

  The ant colony algorithm is a class of algorithms that fall under the label "metaheuristics". These are problems where the computer have to just "look around" for the solution. Try new values, new paths and new combinations to arrive at some solution to a problem that is a bit or a lot better than the initial solution. As has been proven by the "no free lunch theorem", there's no best strategy for problems in general. Searching for an optimal solution is often likened to trying to find a mountaintop in thick fog. If you walk around enough you will eventually find it, but a common "metaheuristic" in this case is to walk upward from your starting point. Keep doing that and you might reach the mountain top. Or depending on the landscape you may end up on a small hill. The common metaheuristic is then to take note of this location and then go somewhere else at random, try again and see if you get any higher this time. This simple idea is at the heart of many fancy-sounding algorithms including some ant colony optimization algorithms, genetic algorithms, simulated annealing and more.  If any particular wisdom is to be extracted from metaheuristic algorithms, it's probably "don't get stuck in the same chain of thought, take a step back, approach it from a different angle" and "discuss with others, compare notes, see if a combination of your ideas could be better than each alone". This is not terribly deep, but a lot of metaheuristics are born out of common-sense thinking, so it's really the computers learning from us and not the other way around.