The appeal of Machine Learning (ML) lies in the idea of computers teaching themselves to solve problems, rather than relying on humans to specify their every move. Relying on humans to hard-wire behaviour is limiting because of the obvious difficulties of anticipating any number of situations in advance, particularly in a changing world. But further, we often simply do not know how it is we do what we do, and so cannot specify it to a computer. And, of course, there are any number of problems we have trouble solving ourselves in the first place. So there are limits to the complexity of the problems we can address by building hard-wired solutions.
KeywordsDesign Space Reinforcement Learn Classifier System Sequential Task Learning Agent
Unable to display preview. Download preview PDF.