Google has become more powerful than ever with its rapidly enhancing artificial intelligence technology. We all are well familiar with the Google translate but the problem is that it cannot translate everything exactly written in another language. When any line does not make the sense then Google translator change it accordingly. But that translation then does not give us the exact result of the translation. This problem can now overcome with the help of artificial intelligence process used in the AlphoGo. The team of Google has recently developed a computer that beats the most powerful player of the ancient game of Go. There was a match organized between Korean grandmaster Lee Sedol and AlphaGo that was so powerful, so upsetting and so exhilarating that it turned into the cover story for the magazines. According to reports, Lee had won all the matches but it fails in front of the AlphaGo. It shows that a computer can beat the humans and can surpass the human talents, thus artificial intelligence has the ability to pull humans to new heights. Korean grandmaster Lee after playing with AlphaGo said that this machine has opened its heights and show him the new ways of playing this ancient game. The technology that underpinned the AlphGo that is known by the name deep neural networks is now playing a very significant role on Google Translate.
Â
AlphaGo with deep neural networks
Deep neural networks are actually the same breed of artificial intelligence technology that recognizes people in photos posted on Facebook and identifies commands spoken into Android Phones. This technology is modeled on the way neurons are connected in the human brain. According to Google, its new system GNMT or Google Neural Machine Translation is able to reduce the translation errors by 60 percent. For now, it can translate from Chinese to English but in future, it is expected to translate more than 10,000 languages. Not only Google but all the big tech giants are working on the same thing by training deep neural networks using translations gathered from the internet. By moving on this technology all tech giants can improve their machine learning in a much faster and broader way. Google is now using long short term memory technique that itself is a form of deep neural networks. The importance of LSTM or long short term memory is that it can retain both long-term and short-term information. Researchers are now trying hard to use LSTM technology with language translation. Google is currently training its neural networks using graphics processing units. All the big tech companies are now racing towards each other for achieving a better artificial intelligence technology.