Chapter 10. Reinforcement Learning

In Chapter 3, Learning from Big Data, we were introduced to two fundamental types of machine learning techniques: supervised learning and unsupervised learning. In case of the supervised learning, a model is trained based on the historical data (observations) for predicting the outcomes based on the new data inputs. In the case of unsupervised learning, the model tries to derive patterns within the datasets and define logical grouping boundaries in order to separate the solution space. There is a third type of machine learning algorithm that is equally important for the evolution of artificial intelligence.

Remember the process of learning to ride a bicycle. We observe another person who is riding a bicycle, create a mental model on how to do it, and attempt it ourselves. It is not possible to just get the balancing and movement on a bicycle right in the first attempt. We (actor) try for the first time (action) on the road (environment) and may fall down...