How does RL relate to genetic algorithms?
Most work with genetic algorithms simulates evolution, not learning during an individual’s life, and because of this is very different from work in RL. That having been said, there are two provisos. First, there is a large body of work on classifier systems that uses or is closely related to genetic algorithms. This work is concerned with learning during a single agent’s lifetime (using GAs to organize the components of the agent’s mind) and is in fact RL research. The second proviso is that GA work is often related to RL by virtue of being applied to the same problems. For example, GA methods can be applied to evolve a backgammon player and that player can be compared with a player learned by RL methods. In fact, a large portion of systems evolved by GAs are controllers that could alternatively be learned by RL methods. It is tempting here to make a blanket statement about which class of methods is more appropriate or performs better. A crucial distinction is that between problems in
