Machine learning from the market point of view
I was asked in a recent interview if I had any knowledge in Machine Learning, and, for a certain moment I just replied: “I have some idea”!
The interviewer was soon surprised, as the position was precisely for a Specialist in Machine Learning. Then, I tried to explain! I say that I am aware precisely because I understand that the next steps of the new technological evolution keep something that what we know today is mere learning. Although all this evolution is the result of professionals in this area, the vast majority know the superficial. Great minds, geniuses, science revolutionaries, mathematicians and philosophers, perhaps have a better perception of the concept! I do not say this by questioning our abilities, but to remember that Machine Learning has more to do with these topics than we think.
Perhaps Machine Learning in the conception of market, is to know more about the main prediction models, to understand the cases and their applications, to know how to test them, to understand their philosophy in the face of problems, how to structure them in the chain of procedures, to know when to use, organize training samples and the best infrastructures for certain situations, how accurate it is. Well, this is not a small think, I know. However, understanding that it is only one aspect beyond a huge amount, is fundamental. In summary, through the “flute of life”, we (a large number of computer scientists, engineers, or anyone else who does focus on market) decided to follow a path that shows an intelligent and long piece from a magnificent chapter of technology that is the art to understand our biological behavior, but, of course, not unique.
The crucial thing about Machine Learning (…and what fascinates me most) is the fact that at some point man will not decide on machines and, for that to happen, man will need a really high understanding of philosophy, sociology, mathematics and countless other areas of knowledge, in addition to computing. To give an example, a old problem: “a self-driven car is in a highway, carring one little kid. Speed limit is ok, but suddently you are just facing 4 other kids crossing the way. What would be the best choice for this self-driven car?”
Evolution is imminent, and that’s might be a good thing! Whereas, the evolution of machines will change the basis of computing! Those of mid-1942 until 60’s, with the theories of Alan Turing and others, regarding the creation of finite automata and machine automata. Today, we live to definitely see the breaking of that barrier (if it hasn’t already – The Loebner prize awarded participants at most, with bronze medals - the silver and gold stages were never reached by the machines).
Loebner Prize gold medal
I think we are about to witness a reformulation of machine states (finite states? Are you sure?). The computational power and dimension of the instruments available today will allow this. The synergy between the multiple areas of knowledge are much closer and closer, easy to conclude that literacy will soon serve another domain unless the one we usually know. I can be totally wrong, but I truly believe that in some aspects the technology might be following some of these steps. One thing is for sure, this is a long discussion and, more and more, I have in mind that I only really have an idea about machine learning.
Finally, I would like to share links that I use from time to time to update myself. In these links, there are discussions and many articles rich in information, in addition to being at the forefront of development in terms of machine learning.