chapter  4
Combined Learning
ByHamidou Tembine
Pages 57

In Chapter 2, we have seen various classes of strategy learning schemes. In Chapter 3 we have provided payoff-learning schemes. In this chapter, we will go one step further by combining the two approaches.

Distributed schemes may have local information but may also have uncertain knowledge of the system parameters. The challenge is to design fast, convergent, fully distributed learning algorithms that perform well even in the presence of errors (noises) about the system measurements. As we will see, information is crucial in the process of decision making.