Continuing next in my series “Data Science - A Primer”
(Kindly read it too if you haven’t until now)
So far, we now know what is Data,what is Science and what is Data Science.
We have loads of data.
Do you remember the population explosion problem we should be worried about?
With increasing population, data increases manifold.
Just think about it. How easy it would have been in the times of Adam and Eve. Just two entity, just one area, few countable objects to render. Such less data.
But today data has increased exponentially, and I am not exaggerating, nor I am being paranoid but I do need a glass of water thinking about all this now.
Furthermore account this for the fact that ever since corporations have understood the importance more-so monetary benefits of Data, they are ever so involved in collecting it.
(So much so that there is a special branch of data science for effective collection of data)
What to do with all this data?
(I would post an image but I am running out of time here, Sorry! Just imagine Leonardo di Caprio with extended Hands)
I mean it is officially out of my human bounds.
I cannot even sort this data as and when required manually let alone performing typical operations on it.
This is where Machine Learning comes into play.
Machine learning is the branch of Data science that has the job of finding simplest model that fits the given data, and based on that model we can then deduce cool things about our data, even somethings which weren’t already there.
Machine learning requires knowledge about data. There dependability on each other and so on.
What if I do not know everything about data?
What if we have loads of data, but we do not know algorithms or models that support it for effective extrapolation?
What if even if such a model exists it is too complicated for us to fathom?
What is we do not even care?
We just care about the magical part.
We supply the data in, we get the information out. (This is typically how I was taught in elementary school, never knew they were so ahead of time)
This the era of Deep learning.
We feed monstrous data to machines.
Then we wait.
We test them, we “Train” them.
Machines “learn” the “Patterns” in data.
Instead of knowing these “Patters” or “Dependability” or “Relation” in data we now have a black box.
We feed the Data in, We get the information out.
This is how we got from Data science to Machine Learning to Deep Learning!