Vehicles that can be driven on their own, assistants who are capable of translating from one language to another, or personalized purchasing suggestions. Complex tasks that used to be seen as a fantasy, today can be possible due to 'Machine Learning', a quality that allows computers to learn by themselves to perform functions without having to be programmed to do so.
What is Machine Learning?
It is nothing more than an artificial intelligence discipline that provides computers with the ability to see massive patterns through data and thus be able to make their predictions. This is what allows those computers to be able to perform functions autonomously, that is, without being programmed by anyone.
Although this term has been heard since 1959, it has only become relevant in recent years due to the constant increase in the power of computing and data. The techniques used in automatic learning are a fundamental pillar in the process of Big Data.
Three categories divide the Machine Learning algorithms, although the most common are the first two.
Supervised learning: this algorithm has the support of a previous learning that is based on labels and data that allow you to make your decision or make a prediction.
Unsupervised learning: This type of learning is totally lacking in previous experience or knowledge. It takes care of shaping a whole set of data in order to get a pattern to organize the data correctly.
Learning by reinforcement: The algorithm must learn and gain knowledge through their own experiences. This implies that it must take the decisions it considers correct before different situations and all this is achieved through trial and error.
Automatic or autonomous learning
It can produce behaviour through algorithms, and these at the same time are supplied by a large amount of data. The algorithm with all this will know what decision to make. From there the machine will be able to automate the tasks and functions.
Today's technologies are increasingly robotic. Some time ago was launched Autonomous Database, this automates the data using machine learning algorithms.
Also known as Deep Learning, it tries to capture concepts more efficiently by analyzing data in an abstract way. It does this by applying compression of the non-linear. It works in a similar way to the brain. Each neuron can accumulate layers of data in order to learn the concepts.
In order to have an idea about the functioning of DL, let's suppose that the neuron network wants to recognize faces by forming the face through layers that will give one by one the contour of the face until being able to distinguish specific faces.
Some applications of Machine Learning
Social networks: Facebook, Instagram, and Twitter use learning algorithms, Twitter to detect existing spam, and Facebook to detect false or disallowed news in broadcasts and live.