Reinforcement machine learning

The term reinforcement means "the process of encouraging or establishing a belief or pattern of behavior" or"positive feedback leads to reinforcement".

So reinforcement machine learning works on a reward base model.

The main approach is to maximize reward in a particular or specified situation. 

The main functionality which differs from Reinforcement learning and supervised learning is that in supervised learning the training data has the answer (output) with it so the model is trained with the correct answer (output) itself whereas in reinforcement learning, there is no answer(output) but the reinforcement middle man or sometimes known as agent decides what to do to perform the given task or next task. In the absence of a training dataset, it is free to learn from its past and upcoming experience.

 

Example: Chess game.

Main Types of Reinforcement: There are two types of Reinforcement models which are mentioned below.

 

1-Positive Reinforcement :

Positive Reinforcement is defined as when an event, occurs due to a particular behavior, increases the strength and the frequency of the behavior. In other words, it maintains a positive effect on behavior.

 

Advantages 

• • Gives Maximum Performance
  • Maintain the same state every time for wrong input

Disadvantages 

• • Too much Reinforcement the model can lead to an overload of states.

2-Negative Reinforcement:

 Negative Reinforcement is defined as the strengthening of behavior because a negative condition is stopped or avoided and punished on time to time.

 

Advantages

• • Increases model behavior and accuracy
  • Provide resistance to the minimum standard of performance

Disadvantages

• • It only provides enough to meet up the minimum behavior
  • Too much negative reinforcement of the model can lead to an underload state.

 

Applications areas of Reinforcement Learning 

• Industrial automation.
• Data processing
• Training systems