So, today it just occurred to me that if ML is a branch of AI, then are there any more branches and if yes then why is that ML is so popular.
I figured that there are other branches as well that help us make machines more intelligent than simply us having to cover all the real world scenario as rules in our code- which is practically obv not possible for us to do in every case. I mean, just imagine having to code the rules for identifying a cat!
Now that is exactly why ML became so popular over other branches. Other branches for example- Expert Systems, Planning and Search require humans to hard code rules or the rules are finite (as in a chess playing software). Some common problems with translating real world scenarios to computers as rules-
situations are open ended- no finite rules possible every time.
lengthy manual process- extract all the possible info from the minds of domain experts, make sure the edge cases are covered as well, translate them all as rules in code.
as the list of rules increases, maintenance becomes a nightmare with potential conflicts too because as we understand, our real world hardly ever has two situations as completely same. there is so much of additional context to it that alters the rules that must apply to it.
So, basically ML helped solve our problem of- trying to code our common sense into code! It enabled machines to learn from data and come up with rules themselves with decent accuracy as well.
No responses yet.