public void Step(double reward, int nextState)
{
var nextAction = ExplorationPolicy.SelectAction(_q[nextState]);
var target = reward + DiscountFactor * _q[nextState][nextAction];
var delta = target - _q[CurrentState][SelectedAction];
_q[CurrentState][SelectedAction] += LearningRate * delta;
CurrentState = nextState;
SelectedAction = ExplorationPolicy.SelectAction(_q[CurrentState]);
}
public void Step(double reward, int nextState)
{
if (!_visited.ContainsKey(CurrentState))
{
var actions = new HashSet <int>();
actions.Add(SelectedAction);
_visited[CurrentState] = actions;
}
UpdateQ(reward, nextState);
Plan();
CurrentState = nextState;
SelectedAction = ExplorationPolicy.SelectAction(_q[CurrentState]);
}
public void Step(double reward, int nextState)
{
var bestNext = _q[nextState][0];
for (var i = 1; i < ActionCount; i++)
{
if (_q[nextState][i] > bestNext)
{
bestNext = _q[nextState][i];
}
}
var target = reward + DiscountFactor * bestNext;
var delta = target - _q[CurrentState][SelectedAction];
_q[CurrentState][SelectedAction] += LearningRate * delta;
CurrentState = nextState;
SelectedAction = ExplorationPolicy.SelectAction(_q[CurrentState]);
}
public void Begin(int state)
{
CurrentState = state;
SelectedAction = ExplorationPolicy.SelectAction(_q[CurrentState]);
}