What are the areas/algorithms that belong to reinforcement learning?

TD(0), Q-Learning and SARSA are all temporal-difference algorithms, which belong to the reinforcement learning area, but is there more to it?

Are the dynamic programming algorithms, such as policy iteration and value iteration, considered as part of reinforcement learning? Or are these just the basis for the temporal-difference algorithms, which are the only RL algorithms?


3 Answers 3


The dynamic programming (DP) algorithms like policy iteration (PI) and value iteration (VI) are often presented in the context of reinforcement learning (in particular, in the book Reinforcement Learning: An Introduction by Barto and Sutton) because they are very related to reinforcement learning algorithms, like $Q$-learning - They are all based on the assumption that the environment/problem can be modelled as a Markov Decision Process (MDP).

However, PI and VI require that the transition model and reward functions of the underlying MDP are both known. They can also be referred to as planning algorithms, because they can be used to find a policy (which can be thought of as plan) given this transition model and reward function of the MDP, without having to take random actions in the environment. They just exploit the given rules of the environment.

On the other hand, RL algorithms do not usually require or use the MDP's transition model, although they still require the reward function, which is fundamental. They attempt to find a policy or value function (from which the policy can be derived) by randomly interacting with the environment.

There are several categories of RL algorithms. There are temporal-difference, Monte-Carlo, actor-critic, model-free, model-based, on-policy, off-policy, prediction, control, policy-based or value-based algorithms. These categories can overlap. For example, $Q$-learning is a temporal-difference (TD), model-free, off-policy, control and value-based algorithm: it is based on an temporal-difference (TD) update rule, it doesn't use a model of the environment (model-free), it uses a behavioural policy that is different than the policy it learns (off-policy), it is used to find a policy (control) and it attempts to approximate a value function rather than directly the policy (value-based).

  • $\begingroup$ Can you further explain your last sentence ? are you talking about variations like TD[lambda] or policy gradient methods? or something else? Also do you have any recommendation of where I can further read about RL, knowing I already explored the Barto and Sutton book? $\endgroup$ Commented Apr 30, 2019 at 14:59
  • 1
    $\begingroup$ @MiguelSaraiva I have updated my answer. I would recommend that you read that book again and more carefully (and that you start implementing some of those algorithms to get full understanding of those concepts). This is a decent book. However, RL comprises a lot of concepts and dense terminology, which often confuse beginners. Have also a look at this question: ai.stackexchange.com/q/6997/2444, in particular, my answer: ai.stackexchange.com/a/7005/2444. $\endgroup$
    – nbro
    Commented Apr 30, 2019 at 15:19
  • $\begingroup$ I have seen some of the videos in that series by Silver already and they are indeed good. Thank you for your help. $\endgroup$ Commented Apr 30, 2019 at 15:54

In Reinforcement Learning: An Introduction the authors suggest that the topic of reinforcement learning covers analysis and solutions to problems that can be framed in this way:

Reinforcement learning, like many topics whose names end with “ing,” such as machine learning and mountaineering, is simultaneously a problem, a class of solution methods that work well on the problem, and the field that studies this problem and its solution methods. It is convenient to use a single name for all three things, but at the same time essential to keep the three conceptually separate. In particular, the distinction between problems and solution methods is very important in reinforcement learning; failing to make this distinction is the source of many confusions.


Markov decision processes are intended to include just these three aspects—sensation, action, and goal—in their simplest possible forms without trivializing any of them. Any method that is well suited to solving such problems we consider to be a reinforcement learning method.

So, to answer your questions, the simplest take on this is yes there is more (much more) to RL than the classic value-based optimal control methods of SARSA and Q-learning.

Including DP and other "RL-related" algorithms in the book allows the author to show how closely related the concepts are. For example, there is little in practice that differentiates Dyna-Q (a planning algorithm closely related to Q-learning) from experience replay. Calling one strictly "planning" and the other "reinforcement learning" and treating them as separate can reduce insight into the topic. In many cases there are hybrid methods or even a continuum between what you may initially think of as RL and "not RL" approaches. Understanding this gives you a toolkit to modify and invent algorithms.

Having said that, the book is not the sole arbiter of what is and isn't reinforcement learning. Ultimately this is just a classification issue, and it only matters if you are communicating with someone and there is a chance for misunderstanding. If you name which algorithm you are using, it doesn't really matter whether the person you are talking to thinks it is RL or not RL. It matters what the problem is and how you propose to solve it.


It seems that another rather controversial point is about the inclusion of evolutionary algorithms as Reinforcement Learning ones. Sutton & Barto do not. They argue that

Evolution vs Learning Sutton&Barto 20202

And also:

More evolution vs learning.

Other people related with the subject, as the HSE University that offers a course in Coursera, Maxim Lapan , or P. Palanisamy (both Packt's authors) include them into the subject. Apparently they support the idea that RL is defined by changes in performance resulting from interaction with the environment. For instance, Lapan classifies the cross-entropy method as model-free, policy-based and on -policy.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .