Dueling Double Deep Q Learning with Tensorflow