Introduction of Machine learning (Why,How,What)

Before we understand machine learning, lets talk about why machine learning is needed and why the heck we care?

if I asked you to write a program of adding two numbers , you would probably write it but what if i asked you to do multiplication with the same program??

does it work?? Absolutely “No” , that program is written to do addition only

Another example is if I told you to write logic for the below block , how many programs would you write??

1 + 1 = 11        1 + 1 = 10        1 + 1 = 3      
 2 + 2 = 22        2 + 2 = 20        2 + 2 = 5
 3 + 3 = ?         3 + 3 = ??        3 + 3 = ???

You need to write a separate program / logic for every block , wouldn’t it be cool if I had only one program which could solve all 3 blocks? Yes it would.

That is where machine learning comes into picture , No need to write programs with hard coded rules and let the system understand the logic and produce the desire results.

Hence Proved.

Now lets about how??

The below picture explains clearly

So instead of giving the program / logic, we are giving the output

For example , Addition of two numbers , for traditional programming we give data and logic, a=2 , b=3 then we get answer 5

For machine learning we give a=2, b=3 and output = 5 so the system understands it ,how come 2 , 3 is 5 ?

Data for addition task is below

Addition   Above examples(logic)
2,3 5       1,1 11      1,1 10    
3,4,7       2,2 22      2,2 20
4,5,9       3,3 33      3,3 30
.....       ......      ......

And finally we ask something like what is the result for 5,6

we get the output as 11 ( actually it might be 10.989 or 11.01 etc.. depends on the data how much we provided).

“The more data we have the better result we get”

So What is Machine learning???

Machine learning is all about learning from data(examples) , building the logic and predicting the output for a given input

A definition from Tom Mitchell( A famous computer scientist) is this

“A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P, if its performance at the tasks improves with the experiences”

so we can break that down into 2 parts

1.Learn from past experiences

2.Improve the performances of intelligent programs

That’s it for this Story , Hope you get some idea.

In the next story I would like to talk about different types of machine learning(what,how ,why)

See ya!