With this series of Posts, I’ll dive into the currently hot topic of machine learning and log my explorations with the topic. This post will act as an introduction to my explorations and the way forward that this series will take.
When I first heard about machine learning, it sounded like magic. It seemed we could just simply take data and feed it to computer like food, and just like that the computer starts doing whatever we want it to!
Of course this was a very naive view of ML, but it’s fundamentally the same. We create a function (say f), and we try to optimise this function by feeding it data, and the resultant approximation of this function should do what function f was supposed to do.
If that isn’t very clear to the reader, it’s nothing much to worry about. I mentioned it briefly in my previous post on AI and will elaborate on the mathematical portions in the upcoming posts. What’s more important here is, our job is not done once we have data. We *need* to define the function f and we *need* to define how optimisation will happen. These definitions come from mathematics, and require some probability and statistics to appreciate fully. But in terms of coding, we don’t really need to know all this, we have several libraries that have worked very hard to conceal this and we just need to provide them our data.
The remainder of this post will focus on what Machine learning can do, and hopefully provide an insight to the reader from my perspective of the field, and what are the possible areas that they might want to explore. Essentially the what and the why and ML.Continue reading “ML101: The what and why”