Let machine do the learning
Rather than the fancy name, machine learning is actually an idea with history — No, it’s not like thousands of years. The subfield of computer science gains popularity due to rapid improvement of hardware and software in the modern world. It becomes a powerful tool not only for A.I. development but various fields requires insane amount of calculation.
Machine learning is literally how machine learns the pattern from data and let it do the repetitive work for you.
The gist is to let the computer calculate how bad the model is, and adjust the parameter accordingly. Only catch is it has to be iterated many times. Human gets bored easily and makes error, but computer follows through. That’s why we let computers do the work.
It may not sound very efficient or smart, and by the reason it wasn’t used widely for years due to the computation capability. Nonetheless, it’s a practical way to approach the best performed model.
In order to represent and approach the question in a way computer can follow through an algorithm to find the best solution, there is a lot of statistics and calculus involved. Thanks for the MOOCs and many open learning platform, it is quite easy to acquire those knowledge and brush up on that. To implement the model, some coding skills are required.
I might jump around starting the next article given there are many interesting research or application coming out almost every week, and I will also document part of my projects to be slapped by the master passing-by.