CQLearner Class Reference
Class for Q-Learning.
More...
#include <ctdlearner.h>
Inheritance diagram for CQLearner:
List of all members.
Detailed Description
Class for Q-Learning.
Q-Learning chooses always the best action for the state s_{t+1},
which doesn't have to be the action executed in the state s_{t+1},
since exploration policies might choose another action. So
Q-Learning is Off-Policy learning, it doesn’t learn a the
values for the agent's policy, but for the optimal policy.
The class is just a normal TD-Learner, initializing the
estimation policy with a CQGreedyPolicy object.
Constructor & Destructor Documentation
|
CQLearner::~CQLearner
|
( |
|
) |
|
|
The documentation for this class was generated from the following
file:
|
