Why do we need machine learning?
That’s the first question I asked myself when I first met the topic. We, humans, are already capable to read, understand and extract conclusions of almost everything in the world. And we also hope to give helpful advice if we are expert enough.
Then, why? Because we are just that, humans. Sometimes there is something we hadn’t seen, there are challenges we are not prepared to success in. Because there are a million things that could be wrong. And because, face it, we are expensive enough to think of something to at least help us do the same but faster, cheaper, more accurately.
On the other hand, I never said that machines would replace us and win a war against us. We are still as far from it as reaching Pluto and living there as we live in New York. But they can be of great help today.
There are two main kinds of problems to be solved in machine learning: classification and estimation.
- Classification is when the system says if the data belongs to one specific class, or another.
- Estimation is what is going to be the probable outcome of the input, unseen data.
The problem is predefined in both cases.
Many other problems can be proposed, but in the end, it all comes of these two old guys. In the following posts, I’ll cover some solutions to classic and novel problems, but feel free to propose your own, and we’ll try to solve them together!!
If all places in the universe are in the Aleph, then all stars, all lamps, all sources of light are in it, too.
Jorge Luis Borges, The Aleph