On Wednesday, Microsoft Researchers announced some impressive gains in machine translation that resulted in a Chinese to English accuracy rate on par with an actual human to human translation.
With several 2,000 sentence tests run and confirmation from 3rd party bilingual language consultants Microsoft Researchers had enough data to confidently attest to the accuracy of its new machine translation.
“Hitting human parity in a machine translation task is a dream that all of us have had,” said Xuedong Huang, a technical fellow in charge of Microsoft’s speech, natural language and machine translation efforts.“We just didn’t realize we’d be able to hit it so soon.”
To attain the record speed at which this achievement was gained, Microsoft Researchers used dual learning for fact-checking translations and deliberation networks to refine translations. Microsoft Researchers also combined the aforementioned techniques with new ones that include boosting English-to-Chinese and Chinese-to-English translations with joint-training as well as agreement regularization to mimic left-to-right and right-to-left sentence reading.
Microsoft Research has aspirations for the potential of this new machine translation system that involves improving the field of Artificial Intelligence research, but before soaring to great fanciful AI heights, the MTS will need to be first tested in real-world scenarios.
To gain as many real-world test metrics, Microsoft Research has begun seeding the new translation system here.