[ad_1]
The history of Google has long been linked to the development of artificial intelligence applied in the most varied fields. The famous Translator service, launched in 2006, has also come to implement deep learning and artificial intelligence to deliver more and more accurate service.
Originally, the model of Google Translate created by Franz Josef Och overcame the grammatical rules-based approach to marrying a system based on the Statistical badysis. Thanks to statistical machine translation Och, three years before the launch of the Google translator, he won the Darpa contest .
In 2016, Translate implemented artificial intelligence, via Google's Neural Automatic Translation (Gnmt): a system based on language construction examples from which the "machine" can "inspire" "and, consequently, provide more natural translations .
The new approach makes it possible to translate "sentences" together at once, rather than [tradurre] piece by piece ", reads in the presentation of the company," this broader context is exploited to adopt the most relevant translation, which is then rearranged and adjusted to make it more similar to human speaking with the correct grammar. "
One of the latest developments in the translation service, announced in June 2018, is the possibility of download dictionaries to use translations even in offline mode Despite the advances, machine translation is revealed in many circumstances, especially in the more complex, because the levels of traditional translation entrusted to human sensitivity, n 39 have not yet been achieved (safeguard the human factor of communication for example, the value that has been defending for years Global Voices l this translation chosen by the largest companies in the world)
For Alphabet Google's parent company, 2016 C is also a key year for another aspect of the applications of intelligence artificial: it's the born Waymo, the active division in the development of autonomous cars.
Really the experiences in the automotive field had already begun in 2009, under another name (Google Self-Driving Car Project), but from 2016 an agreement with FCA has heralded the first commercial model of cars Google based on the "base" Chrysler Pacifica. Today Waymo is the only company to have made a fleet of 100% autonomous cars available (located in Phoenix, Arizona).
"The artificial intelligence plays a crucial role in almost every part of our car system" guide ", written last May Cto di Waymo, Dmitry Dolgov " while the perception is the most mature area for deep learning, we also use deep networks for everything: from prediction to planning, mapping and simulation.With machine learning we can guide in nuanced situations and difficult to maneuver in areas under construction, give way to emergency vehicles and leave space for cars in parallel parking. "
Again, here are the" examples "with which RNs were "trained" to provide them with the tools similar to human learning
So it is true that deep learning is over in addition established, it is the human communication, on the basis of which is the ability to relate to the context, to act as a guide in this process of refinement of artificial intelligence. A model, the human model, which is fundamental for the understanding of the message and therefore of learning, in the sectors of translation and the automobile .
As proof of the importance of an empathic communication, you can listen to the audio translation of this article, also available in English by clicking on the following link .
[ad_2]
Source link
