What is Machine Learning

The machines are capable of learn. But don't let this keep you awake with apocalyptic visions of a world dominated by robots that have enslaved or exterminated humans after “learning too much”. And let's not digress, either.

The machine learning consists of a discipline of computing in which, and through artificial intelligence algorithms, computer systems they are able to simulate the human learning process, solving situations and challenges for which they had not been programmed previously.

Historically, computers have been unable to do anything other than what they were programmed to do. but the gradual increase in power in the hardware has allowed to take them further, accompanied by the algorithms of software.

How is a machine capable of learning?

The learning process of a computer system is carried out based on the analysis of large amounts of data.

What the software does is, broadly speaking, build predictive models from the analysis of the data, although not modifying its own programming, something for which the machines are not yet ready.

However, in a certain conceptual way, we can consider that the building of models on which to act, it is a kind of modification of its base programming, although this at the level of code source is not really like that.

Machine learning is a branch of the discipline of artificial intelligence.

To finish, let's put some examples of the use of machine learning, starting with an investment system in the stock market.

This one is capable of analyze the rises and falls in stock prices, so that, without having been previously programmed to predict a certain scenario, through the analysis of a large number of factors (prices of the shares of other companies, comings and goings of the market, investments made by others, ...), you can calculate when it will be more conducive to buy and / or sell certain securities.

And, what is usually more important in these systems is that they are able to analyze their own performance and "learn" from their successes and mistakes, "improving" their "performance" over time.

In fact, I have not chosen this example because, but because today most of the operations in the stock market are carried out by computer programs of this type.

We can enjoy the benefits of machine learning in a streaming music service like Spotify.

Is it true that when we listen music In Spotify (or, eventually, in other online services of this type), does the program recommend other groups or songs according to what we listen to? And is it not less true that these recommendations evolve as our habits of consumption of music in the toilet?

How do both Spotify and the other online music services do this? Simple: with a machine learning system that learns what we like, and decides what to recommend based on it.

And the more music and the longer we listen to, the more the system will learn of our preferences and, therefore, the better chance it has of hitting its recommendations.

Fotolia photos: Aleutie / Kit8

Topics in Machine Learning