Intermediate · Reinforcement Learning

Policy

Visual diagram · (in preparation) · Math · (in preparation) · Worked example · 3 difficulty levels.

TL;DR. In reinforcement learning, the strategy an agent follows to choose actions given a state.

Technical Definition

In reinforcement learning, the strategy an agent follows to choose actions given a state.

How it works

A policy π(a|s) maps states to a probability distribution over actions. Deterministic policies pick a single action; stochastic policies sample. The goal of RL is to find an optimal policy that maximizes expected cumulative reward. Policies can be represented as lookup tables, parameterized functions, or deep neural networks (deep RL).

Related Concepts

• Reinforcement Learning — A paradigm where an agent learns to make decisions by receiving rewards or penalties from its environment through trial and error.
• Reward Function — A scalar signal that tells a reinforcement-learning agent how good its action was.
• State / Action — The two core elements of every reinforcement-learning step: what the world looks like and what the agent does.