Humans are very clever and we need very little information to acquire knowledge. Now, an algorithm is able to imitate us to learn to be creative.
Researchers at Dewp Mind, Google, (artificial intelligence Company) have developed a connective reasoning algorithm that has succeeded in overcoming humans in a complex imaging test, that is, a relational reasoning in computers being very similar to the neuron system that is found in the human brain.
Then… WHAT DO YOU PROPOSE?
The study proposes a new way: the artificial neural network capable of relating concepts! Just as neurons are connected to the brain. Artificial neural networks are assembled together through small programs that find patterns in collaborative data.
They have specialized structures for processing images, analyzing language, or even learning games. Based on some tests of relationships already made in this study, they were already overcome by the system.
It obtained 96 % of positive results over 92 % obtained by the people who participated in the experiment. It exceeded the capacity of human reasoning.
It also excelled in language exercises, particularly in inference exercises where his success rate was 98 % versus 45 % for other AI programs.
The relational reasoning is a normal behavior in people, but when we refer to computers or machines, thus, through statistics or learning, are good for the recognition of patterns already established but not for the use of the logic.
This new network of Artificial Intelligence (AI) relationship is designed to compare pairs of objects in different scenarios individually.
So researchers are explicitly forcing the network to find the relationships between objects, thus challenging to the network to carry out various tasks such as: answering questions about the relationships between objects – cubes, balls, cylinders – that appeared in the same image.
The ability to reason, relate and create connections are basic aspects of the sophisticated thinking that Artificial Intelligence (AI) had failed to manage.
This being is a breakthrough, but according to experts, to achieve the flexibility of human intelligence, the network should learn, for example, to compare not only pairs of things, but trios, double pairs and other combinations.
Artificial Intelligent computers are a new method that is taking more space and public in the field of the technology positioned as one of the most important aspects with the many benefits of the world of technology.
Read Also: Top 6 New Technologies of 2017