When I entered the Polytechnic School in Brazil, back in 2009, I didn’t have a good idea of what an “engineer” was. I just wanted to learn more things and somehow work with mathematics, my favourite subject in high school.
After 3 years and a half studying there and almost a year here in the US, lots of things have changed. I chose control engineering as my major, I learned how to program and was exposed to (sometimes tortured by) various other subjects.
I will have two more semesters before graduation when I come back to Brazil, but only if my final project is done by July/14. So I should start it now if I don’t want to wait for December/14 to finish college.
There’s just one Big Problem: I need a project idea. Not that there isn’t any — I just don’t know which one. After some research and lots of papers read, my interests boil down to artificial intelligence and probabilistic models.
Fancy terms, I know, but basically I want to build a machine learning system that’s complex enough for me to write a Master’s thesis about it, but simple enough that I can finish it during these 2 semesters without suffocating myself. This balance is hard (and future forecasting is impossible).
So far, two themes have been robbing my attention: natural language processing and autonomous robots. I have almost no knowledge of the former, but it seems more likely to be useful for data analysis today. The former is the subject of a project I’ve worked on in the laboratory of automation and control during my sophomore and junior years — I know more about it, but it’d be impossible by myself, as I’m not proficient in hardware.
Besides, it should do something, not only show “hey, look, I know NLP and how to program, cool huh?” but actually be useful. It might turn into a commercial project eventually, but initially I prefer to think of it as a complex engine for a web/mobile application. Something I could ask for feedback on Tumblr, Twitter or whatever.
For the moment, I decided to continue meditating on it and simply learn about libraries and theories that can be helpful — that’s why I chose Natural Language Processing and Probabilistic Graphical Models as my two next Coursera couses.
I still have lots of work to do. (…)