What is Reinforcement Learning? Free Tutorial

REINFORCEMENT LEARNING

Here in reinforcement learning, we are  going to have an environment in which the Agent or AI is going to be left in. Let's consider a maze.

This is How it Works

The agent (AI) in the environment causes an action leading to the change of the state(current condition) of the AI . It either goes close or away from the desirable result.This change of state leads to a reward.If the Agent goes closer it gets a award and in case it moves away it will get a negative reward .This process Happens repeatedly and it becomes aware of where it should go depending on the reward.

Lets take an example from daily life,Let's consider you are cooking omlette.At first you put the flame on,then wait for 1 hour and pour the egg(it will burn a negative reward).Next time you try doing the same faster and you get a positive reward .

Here is another example for you:-

A dog is begin trained.It's commanded to sit and it doesn't react.It get.'s a negative reward (gets hit or no treat)The next time it sits for the same command and it sits down.It get a treat(positive reward).

The same happens here but,

it's a 

+1 and-1

digital reward

When something is trained through reinforcement learning it doesn't have a pre-program

on how to do it's work  

The Bot or AI is only given the Goal and Desirable result

When it does not create a desirable result it's given a -1

hence,does not repeat the mistake

When a desirable result is created the AI is given a +1

eventually,from this process it learns how to reach a given goal.

Hope you got the knowledge desired...

More about AI coming every 2 days

Subscribe to notified.