ajout q learning approximatif

reinforcement/qlearningAgents.py

@@ -208,14 +208,25 @@ class ApproximateQAgent(PacmanQAgent):
           where * is the dotProduct operator
         """
         "*** YOUR CODE HERE ***"
-        util.raiseNotDefined()
+        features = self.featExtractor.getFeatures(state, action)
+        weights = self.getWeights()
+        q_value = 0
+        for i in features.keys():
+            q_value += features[i]*weights[i]
+        return q_value
+        # util.raiseNotDefined()
 
     def update(self, state, action, nextState, reward):
         """
            Should update your weights based on transition
         """
         "*** YOUR CODE HERE ***"
-        util.raiseNotDefined()
+        features = self.featExtractor.getFeatures(state, action)
+        weights = self.getWeights()
+        difference = (reward + (self.discount * self.getValue(nextState))) - self.getQValue(state,action)
+        for i in features.keys():
+            self.weights[i] = weights[i] + self.alpha * difference * features[i]
+        # util.raiseNotDefined()
 
     def final(self, state):
         "Called at the end of each game."