When machines learn on their own


Machine Learning works with smart algorithms. A person can write the algorithms and then give the computer a series of logical steps. And, just like children and pups, machines can learn by doing. Different time frames for each direction will make machines to do processes in a more efficient way than the original program.

These techniques may reminiscence popular films, such as I, Robot or Space Odyssey where machines learn how to take control. Reality is much more concrete and focused. Machine Learning is used to select and classify huge amounts of data.

Researchers from Tec de Monterrey have been working in this subject for a few years. In the School of Engineering and Sciences they developed safety apps, wearable bands, and robots that teach autistic children about emotions. This year, they reached the second place on The RedICA Text-Image Matching Challenge.

By using their own knowledge and research they decided that they wanted to predict in an efficient way the relation between a text and an image. In order to be a stronger team, they collaborate with Intelligent Systems researchers. For the online international competition, Tec researchers used the system codalab and the data provided by the organization.

The reason their work stand out among the rest of the teams was because a contrast pattern. This is an initiative that not only has been explored as a research line but has also been internationally endorsed. The proposition of the Mexican institution keeps on gaining track in the technology field.

In the end, worldwide researchers are working not to earn prizes but to build real solutions for society. Teaching machines to do manual work was a key factor to improve life quality during the Industrial Revolution. Teaching machines to handle data will help current and future communities to take smarter decisions. 


Originally published on THE Digital Hub.