Softmax Policies Recall the epsilon-greedy scheme from lecture for balancing exploration and exploitation. In this question, we'll explore a different way of balancing exploration and exploitation. So far, we've seen deterministic policies mapping states to actions. In contrast, a stochastic policy lets the agent randomly choose between multiple actions from a given state, according to some probability distribution. Formally, ( ) (as) represents the probability of action a being taken, given that the agent is currently in s