Contents

## How is a policy used in a MDP?

A policy is a rule for selecting actions For MDPs this can simply be a mapping from states (of the underlying system) to actions updated by Bayesian conditioning after each action For POMDPs this is not possible, because the system state is only partially observable Thus, a policy must map from a “decision state” to

**How are MDPs and state value functions related?**

MDPs introduce control in MRPs by considering actions as the parameter for state transition. So, it is necessary to evaluate actions along with states. For this, we define action value functions that essentially give us the expected Return over actions. State value functions and action value functions are closely related.

### How is the reward received in a MDP?

This reward is received for being at the state S_t. By convention, it is said to be received after the agent leaves the state and hence, regarded as R_ (t+1). An MDP is defined by (S, A, P, R, γ), where A is the set of actions. It is essentially MRP with actions.

**What do you mean by RL agent in MDP?**

First things first, before even starting with MDPs, we’ll quickly glance through the terminology that will be used throughout this article: Agent: An RL agent is the entity which we are training to make correct decisions (for eg: a Robot that is being trained to move around a house without crashing).

## Which is an example of a dominated strategy?

Home Economics Game Theory Dominated Strategy Dominated Strategy. A dominated strategy is a strategy which doesn’t result in the optimal outcome in any case. A strategy is dominated if there always exist a course of action which results in higher payoff no matter what the opponent does.

**What are the payoffs of a dominated strategy?**

In the payoff matrix above, rows show strategies of Firm A and columns show strategies available to Firm B. The numbers in red to the left of the comma in each cell are payoffs of Firm A and those to the right (in blue) are payoffs to Firm B.

### Can a game have more than one dominated strategy?

Since not all games have a dominant strategy, it is not necessary for all games to have dominated strategies. But if there are more than two strategies available, it is possible for a game to have a dominated strategy even if there is no dominant strategy (as illustrated in example 2).