Professor Laura Burian, the new dean of the Middlebury Institute’s Graduate School of Translation, Interpretation, and Language Education, says her priority is to adapt the curriculum to prepare students to lead in a rapidly evolving language and technology market. Continue reading
As Facebook faces fundamental questions about its business model, understanding what content is being shared on the platform has become a top priority. With more than 2 billion monthly active users, language tech will play the decisive role. Continue reading
Facebook is open sourcing its unified deep learning framework PyTorch 1.0, the underlying technology that supports AI and many of its products including Translate. Continue reading
Research into neural machine translation in 2018 is exploding, with April 2018 being the busiest month ever for NMT, as Big Tech companies from Google to Facebook to Amazon go full throttle on multiple areas of research. Continue reading
In first trading update post Moravia acquisition, RWS says revenues jumped by 82% to GBP 139.6m and pre-tax profit was at least GBP 28.3m. While analysts continue to rate the stock a buy, the company lost over USD 250m in market cap in early London trading. Continue reading
Facebook is inviting the academic community to submit research proposals for a USD 20,000 to USD 40,000, one-year research grant starting June 2018. Facebook research will award up to four research proposals and is open to further research after evaluation. Continue reading
Facebook’s AI Research team delves into the neural machine translation black box and figures out a couple of ways to mitigate performance degradation due to intrinsic and extrinsic “uncertainties.” Continue reading
Massive privacy breach at Translate.com, senior managers leaving TransPerfect in droves, RWS buying Moravia, Facebook launching neural machine translation, Pearl Linguistic going bust, and more than two dozen other stories that lit up the Slator website in 2017. Continue reading
Though Amazon announced its open-source, sequence-to-sequence machine learning framework Sockeye would be available to all last July 2017, its researchers waited until December 2017 to file papers into Arxiv so they can test their creation against WMT2017’s best. Continue reading
Facebook experiments with post-editing to improve its neural machine translation system. The QuickEdit system supposedly improves machine output by learning from post-editors by simply rejecting words in a translation that don’t work. Continue reading
Salesforce launches its own neural machine translation model at Dreamforce conference in San Francisco. The model has not been productized yet but AI unit head hints at deployment in 2018. Continue reading
Servers in Iceland. A million words per second. Record BLEU scores. The founders of Linguee launch their neural machine translator DeepL with bold claims. Experts agree that DeepL will challenge other MT players but call the output far from revolutionary. Continue reading
Facebook says it has completed transitioning to neural machine translation models in its backend systems. Factoring in slang, typos, abbreviations, and all shades and nuances of social media posts, will this really result in more fluent translations? Continue reading
CNBC reports that Amazon is looking to offer the translation technology it built for internal use to third parties. The move would be a direct challenge to Google Translate and the many other MT offerings by big tech companies. Continue reading
Mark Zuckerberg looks forward to “making universal translation a reality” by open sourcing a new approach to neural machine translation. NMT expert impressed by faster training times of the new models but cautions against inflated quality expectations. Continue reading
This information will never be shared with third parties. No Spam.