Google announced that they have finished developing the Google Neural Machine Translation (GNMT) or a new technology in artificial intelligence involved in translating Chinese to English.
According to the engineers at Google, "First, the network encodes the Chinese words as a list of vectors, where each vector represents the meaning of all words read so far ("Encoder"). Once the entire sentence is read, the decoder begins, generating the English sentence one word at a time ("Decoder").
Google engineer Mike Schuster said, "We can train this whole system in an end-to-end fashion. That makes it much easier for [Google] to focus on reducing the final error rate."
Wired.com explains GNMT as "modeled after the way neurons connect in the human brain, deep neural networks are the same breed of AI technology that identifies commands." The GNMT is only available for Google Translate in Chinese.
The error rate now is down to 80 percent.
The search engine giant formed a team called Google Brain or a group of engineers to study how the human brain works and develop technology that imitates it at the same speed as a human.
The developers at Google admit that they have a long way to go, but their achievement now is a great leap forward for the search engine company.
Schuster said, "What we have now is not perfect. But you can tell that it is much, much better."
Google's engineers work on an algorithm that is unique to the company. They use networks that consist of layer after layer of mathematical calculations.
These calculations of linear algebra will be the source for the next layer of calculations, which led to the translator's speed narrowing down to 300 milliseconds per word.