How many branches does Machine Learning have and is any other type of Machine Learning except for Statistical Machine Learning? Which type does Neural Network belong to?

Question

Could someone tell me how many branches Machine Learning does have and is there any other type of Machine Learning except Statistical ML? And which type of ML belongs to Neural Network?

Answer 1

Well, it is not that relevant to count the number of branches. There are firmly some types of learnings which cannot be classified as the term “Statistical”. One of the essential types is learning from just one or a few clear instances. 

Let’s say I am going nearby my friend’s house on the way to someplace, and if a very dangerous dog comes and attacks me, definitely I will not follow the same routine the next day.  

Now, this is already fed in my subconscious mind that helps me to rectify few things. This is known to me very well that dogs have a specific shelter in a home, when they are adopted by families, they will live there for years. 

But, on the other hand, let’s say I am almost hit in the head by a meteor, I can analyze that changing my route will not be helpful since meteors fall pretty uniformly all the earth. But I would start using protection for my head. 

The other type of learning that is necessary for us humans is “learning by being told”, which requires zero first-hand training experiences, but it does require the ability to understand natural language.

And then there’s learning by experimenting on things, which has one-shot and statistical branches. Maybe on some occasions, I might try to walk by the neighbor’s house again (maybe carrying a stick or a can of mace), just to see if that dog is still there and be on the safer side. 

So there is well-done research on mechanizing these kinds of learning in AI, though not so much since everyone became fascinated with statistical ML. Many ML researchers are now so focused on statistical learning that they have completely lost track of the idea that there is any other kind — though they use the other kinds constantly in their own daily lives.

Statistical Machine Learning is very essential. Statistics is invented for a very good reason. It is important and definitely plays a vital role in our sensory/motor machinery. But many times you have a very strong signal like that dog, and not so much noise (well, unless you count the growling), and in this scenario, a few data points will do, or even just one very clear one.

By learning from someone else’s experience is arguably the driving force behind all of human civilization, which is cumulative and is not limited to a person’s lifetime experiences. 

Want to learn Machine Learning, check out this Machine Learning Course offered by Intellipaat.

See this Machine Learning Course for more information :

Answer 2

Well, it is not that relevant to count the number of branches. There are firmly some types of learnings which cannot be classified as the term “Statistical”. One of the essential types is learning from just one or a few clear instances. 

Let’s say I am going nearby my friend’s house on the way to someplace, and if a very dangerous dog comes and attacks me, definitely I will not follow the same routine the next day.  

Now, this is already fed in my subconscious mind that helps me to rectify few things. This is known to me very well that dogs have a specific shelter in a home, when they are adopted by families, they will live there for years. 

But, on the other hand, let’s say I am almost hit in the head by a meteor, I can analyze that changing my route will not be helpful since meteors fall pretty uniformly all the earth. But I would start using protection for my head. 

The other type of learning that is necessary for us humans is “learning by being told”, which requires zero first-hand training experiences, but it does require the ability to understand natural language.

And then there’s learning by experimenting on things, which has one-shot and statistical branches. Maybe on some occasions, I might try to walk by the neighbor’s house again (maybe carrying a stick or a can of mace), just to see if that dog is still there and be on the safer side. 

So there is well-done research on mechanizing these kinds of learning in AI, though not so much since everyone became fascinated with statistical ML. Many ML researchers are now so focused on statistical learning that they have completely lost track of the idea that there is any other kind — though they use the other kinds constantly in their own daily lives.

Statistical Machine Learning is very essential. Statistics is invented for a very good reason. It is important and definitely plays a vital role in our sensory/motor machinery. But many times you have a very strong signal like that dog, and not so much noise (well, unless you count the growling), and in this scenario, a few data points will do, or even just one very clear one.

By learning from someone else’s experience is arguably the driving force behind all of human civilization, which is cumulative and it is not limited to a person’s lifetime experiences.

If you want to become an expert in Machine Learning, check out this Machine Learning course offered by Intellipaat.