MACHINE LEARNING-OVERSIMPLIFIED

Last night I was watching the Movie: I, Robot (Yeah, I know that's a bit old but I like it!). And then I stumbled across an Idea.

The idea was that, can I create a computer program that can accurately (too some extent) tell me about upcoming events. 

No, not foreseeing future but about certain selected things that can be predicted to some good extent.

But in order to that, I need to have some things ready beforehand;

1.A good amount of data (past occurrences) on the event I want to predict.

2. A good computer (which I don't have!)

3. Good programming knowledge.

But in order to do all that, I need to know Machine learning.

But what exactly is Machine Learning?

Take it this way, Suppose you are an infant (sorry!), and you want to pick something, say a pencil.

Now you and I (grown-ups) already know how to do that, but for a person who is born only some months ago is like a computer with very little programs.   

Their mind works something like this, 

1. They see someone doing 'the pick-up' action.

2. Then it processes the mechanics and trigonometry(yeah, even if you don't know, your mind does) of calculating the distance, the right time, amount of energy needed and all that.

3. Then it sends out signal which first activates your shoulder then your arms and finally your fingers which starts the action and picks up the pencil.

(Pardon me if I wrote something wrong in these points, I am no Biology-guy!)

Your mind does all that by 'Learning' from environment and from other people.

Now you have to do the same but with the computer.

So you basically have to program the computer with sufficient code on 'how to learn' things.

The Evolution (One that's interesting!)

Have you ever noticed a newborn/younger's action have found that it is not smooth. Well it is because the kid is still learning how to do certain actions.

This is what we call (or at-least, I call) 'Learning on a small stage'.  

You can call it evolution too, if you want.This evolution, results in precise and smooth movements in adult or teenage years of life. 

Enough of the biology class, now lets get back to computers, shall we?

So after programming the sufficient code for ML (I will use ML for Machine Learning from this point forth), you have to start the program with any game/simulator you want to program the ML Bot for(that you made).

At first your program (ML Bot) will be silly but after a decent amount of time, and a good amount of evolution (It is counted in generations); your Bot will be performing desired actions swiftly. 

Now I can some of you gravely confused. 

Your confusion will be out in a few seconds.

Conclusion

Suppose you want to Program an ML Bot that can play chess very cleverly (Almost like a human), then after programming it completely you have to let it evolve for some generations and after that you will have somewhat a good Chess Playing ML Bot.

So hope you guys got what exactly Machine Learning is but what you and I learned today, is just a penny of the real deal.

There are things like Neural Network, I/O, etc.

that were important to be mentioned but since this blog is all about being "Oversimplification"; I will explain these topics in upcoming blogs.

Thank you so much for reading guys, I hope you all enjoyed it! Please Hit that subscribe button, to know when more oversimplified blogs come out and don't ever forget to leave a comment on what you like and how can we improve things.

Meanwhile you can check out my other blogs too!