For the following reinforcement learning algorithms: (i) Q-learning with fixed and 80% exploitation. (ii) Q-learning with fixed k = 1/k and 80% exploitation. (iii)

For the following reinforcement learning algorithms:

(i) Q-learning with fixed α and 80% exploitation.
(ii) Q-learning with fixed αk = 1/k and 80% exploitation.
(iii) Q-learning with αk = 1/k and 100% exploitation.
(iv) SARSA learning with αk = 1/k and 80% exploitation.
(v) SARSA learning with αk = 1/k and 100% exploitation.
(vi) Feature-based SARSA learning with softmax action selection.
(vii) A model-based reinforcement learner with 50% exploitation.

(a) Which of the reinforcement learning algorithms will find the optimal policy, given enough time?

(b) Which ones will actually follow the optimal policy?