algorithms

(Learning) algorithms that can be used to learn policies for agents.

This package is separated from the smartgrid package, as the algorithms could be used for other simulators (environments). Similarly, the smartgrid env could be used with other learning algorithms.

Learning algorithms receive observations and rewards, and must decide which actions should be taken, based on observations, such that the best rewards are obtained.

members:

Modules

model	The base Model, an abstract class to provide a common (standard) API.
naive	A "naive" algorithm that does not learn (random policy).
qsom	The Q-SOM learning algorithm, leveraging 2 SOMs and a Q-Table.
util	Helper classes that can be used by all learning algorithms.