In a bid to overcome language barriers, social networking giant Facebook has announced a new machine learning translation method, claiming it to be nine times faster than other competitors. Artificial intelligence (AI) has been already in place at Facebook for automatically translating status updates to other languages, but the company is making a transition from lab to app, The Verge reported. Also Read - Meta adds new features in Facebook, Instagram for women's safety in IndiaAlso Read - Facebook to now allow more cryptocurrency ads than before; here's why
We re currently talking with a product team to make this work in a Facebook environment. There are differences when moving from academic data to real environments in terms of language. The academic data is news-type data; while conversation on Facebook is much more colloquial, the report quoted Facebook s AI engineer David Grangier as saying. The new machine learning translation technique has not been implemented yet and exists as a research as of now. But Facebook has said that it will likely happen further down the line.
Usually, AI-powered translation relies on what are called recurrent neural networks (RNNs), whereas this new research leverages convolutional neural networks (CNNs) instead, Facebook s AI engineers explained. RNNs analyse date sequentially, working left to right through a sentence in order to translate it word by word while CNNs look at difference aspects of data simultaneously a style of computation that is much better and faster. So translating with CNNs means tackling the problem more holistically and examining the higher-level structure of sentences. The [CNNs] build a logical structure, a bit like linguistics, on top of the text, said Michael Auli, another Facebook AI engineer. ALSO READ: Artificial Intelligence, IoT to help deliver million midday meals in India: Accenture
Facebook noted that the AI community was willing to improve upon the commonly used RNNs for translation a method that has devoured tremendous efforts already. The short answer is that people just hadn t invested as much time in this, and we came up with some new developments that made it work better, Grangier added.