Machine Learning or The Robots are Coming!

Posted by on October 24, 2013

Lately I’ve been working on some interesting topics at work. Right now I am gearing up for a project starting in November that will utilize machine learning, and using what some call collective intelligence.

Skynet Logo“So what the heck is machine learning? Are you finalizing Skynet or your own robot army, David?”

Well, first, let me say that my 3D printer is nearly complete, and should be before this project starts. So the answer to your question about a robot army is a solid: Maybe.

Machine learning however is both more simple and more complicated than it sounds. Machine learning takes in data, usually lots of data, and data that on the surface doesn’t always make sense, and makes sense out of it. Often finding patterns and relations in the data that help the machine/computer, to make specific decisions based on the data.

Many of the algorithms used are actually ones that humans use everyday, if they are statisticians. They are used to make choices in the stock market, suggest movies you might like based on your’s and others’ previous movie ratings, and what you should watch next on YouTube.

Nearly all of the algorithms are understandable, at the start at least, but deal with so many variables, and such large quantities of data, that the computer does what it does better than any human. It calculates. It does a lot of calculation. As I said earlier, statisticians do similar things, but when they use a computer to do it, the results are downright amazing.

So for this project, I am working on finding the right combination of variables that will lead to the best user experience. I know what is a good user experience, but I can’t sit at every user’s computer and tweak their settings to make it right. That’s where the machine learning comes in. I’ll be coding a set of algorithms to look at how fast their computer is, if they are even using a computer or using a phone or tablet, taking into account the user’s preferences, and the client’s needs as well. From those algorithms, the app will then be able to deliver the best user experience that is possible, at that moment.

At that moment is yet another data point, as are the moments in the past. See, as the user continues to use the app, the machine learning algorithm will continue to learn, and to tweak the settings it has available to make the user experience better and better, all without any downtime.

It’s a daunting task to say the least. It’s also a challenging and exciting one. I’m really jazzed to be working on it, and have been diving into white papers, books, and every article I can on algorithms related to statistical analysis and machine learning.

I see so many uses for it, it’s pretty amazing.

As they say in that movie: “Would you like to know more?”