There are 4 common elements of Reinforcement Learning as shown below
- Policy: Function that basically maps situation to action. It basically states, what action an agent should perform at a given time.
- Reward Signal: The reward is the feedback given to the agent for a particular action. Reward signal can be called as a numeric value which environment send to the agent as feedback of any action. The reward is given to the agent according to the positive and the negative actions.
- Value Functions: This is the quantity that quantifies how a particular situation is good for the agent. It can be stated as the future rewards that an agent will get being in a particular state
- Model of the Environment: This is what defines how an environment will behave when a particular action is performed in a particular state. Also, the agent can predict the next reward and state, according to the current state and action.
Fri, 02/18/2022 - 11:44