Reinforcement Learning for Dumb Kids-(agent, environment, reward, punishment, action)

 

Hola, todays article will cover the basics of Reinforcement learning, we will focus on RL basics cause the advanced concepts are tough, and i myself cant explain them.

so RL in nibba terms, is just real life, we try to survive in our surrounding, surroundings are either know to us, or unknow to us, but due the the hardwired instincts we have, we can survive, and not only surviving ,  we have been dominating planet earth for last 10,000 years.

Now what is reinforcement learning, let me introduce some terms for that 

*agents

*reward

*punishment

*environment

"agent seeks reward in the environment and avoids punishment."    

so u get idea right, its based on reward and punishment, but how is it different than machine learning

well we aint working on data, we are interacting with the environment.

some basic facts:

RL is based on environments, so many parameters come into play, basically they variables are infinite.

scenarios are real world.

broader in sense.

objective is to reach a goal.

Process of RL: starts with an intelligent agent who has to reach the goal and he will move to next state in order to maximize his rewards.

so a short summary of RL process is

" to keep moving until he reaches his goal"

soo let me go in detail of the  process: agent uses a policy to act, in the environment, it works efficiently when it acts intelligently based on principles of gain and loss.

Common concepts:

Gamma: used in state transistion and