It’s called “deep learning” and is considered by many as the future of ‘ Artificial Intelligence .This is a technique used to instruct electronic systems processing through the adoption of neural networks, thus enabling the learning of ideas and concepts to inanimate objects. At the moment it is still evolving, but there are positive results, nor the practical demonstrations of what the potential of such a tool.
The neural networks are artificial in fact a field of study of particular interest to the world of engineering, both academic and corporate. Are several groups that have bet in that direction, among them giants like Apple and Google , whose respective speech recognition systems, as well as the Street View service the search giant, standing on its own foundations on machine learning and neural networks. The ultimate goal in all cases is the creation of a software system able to mimic the human brain to absorb and process information to increase their knowledge.
Just the human brain is the reference model to which researchers are inspired to develop new techniques and new ways to educate those from the physical point of view are nothing but simple electronic circuits able to execute instructions from a programmer. Repeating several times specific experiments is therefore intends to make sure that the artificial neural networks acquire a concept, an idea or experience , the from storing in digital form and the draw so that in subsequent experiments can react in a timely manner.
The idea of developing intelligent systems, on the other hand, is not entirely new, but has had a long history of nearly half a century. It is with the information revolution, however, that this hypothesis by simple theory became reality and in recent years there have been important steps forward. Some students of the University of Toronto, for example, have managed to develop a software able to determine what factors are most likely to represent the ideal solution in the development of new drugs , analyzing the links between themselves and the structure of medicinal products already made in the past. Progress in the field of artificial intelligence thus means to improve several other areas, more or less related to information technology, could potentially apply these tools to each field.
But to what extent can push a system which, although intelligent, is always artificial? To this question they tried to answer the researchers IDSIA, a center for research on artificial intelligence active in the city of Lugano, in the Italian-speaking Switzerland. In the laboratories of the institute has been developed a system that can automatically identify traffic signs and in a race with some German experts was able to do better than rivals in the flesh: of 50 000 signals shown, the 99.46% of them has been properly recognized by the software, while the man was unable to go beyond 99.22%.
The AI can therefore open up new windows of opportunity, which can be exploited in many fields. Last month, for example, Richard Rashid at Microsoft explained to the public a voice recognition software that not only was able to fully understand every single word, transcribing speech into digital format, but also demonstrated its capabilities in terms of interpreting,translating all in Mandarin Chinese and simulating the voice of Rashid , who has never spoken that language.
The challenges in the world of artificial intelligence are still many and before a real revolutionary breakthrough will have to wait much longer. The improvements made in the area but let us hope for the future, a future in which, perhaps, the machines will replace humans not only in manual labor, but also intellectuals, being able to make decisions independently and aware.