Learning element:

It helps in improvising an agent by taking knowledge about performance element and feedback from critic.

Performance Elements:

It is main component of an agent like a Neuron in human body.

Problem Generator:

It experiments on performance element by suggesting actions which can generate new instance or experiments.

Critic:

It compares the performance standard with input and based on the comparison. Critic gives feedback to the learning agent.

Let’s take an example of automated car agent, theoretical components of automated car learning agent can be given as follows:

Learning agent: Create objective (learn when to use break, accelerator, etc.)

Performance element has information about the procedure for driving a car (e.g. actions like: Turing with steering, changing gears, stop car with break, etc.)

Problem generator: Trying different routes to travel is the work of problem generator.

Critic: Observes surrounding environmental conditions and gives this information to the learning element(e.g. check if there is a quick left turn across two lanes of traffic, check if there is speed breaker).