Collaborating Authors

Microsoft improves its AI translations with Z-Code – TechCrunch


Microsoft today announced an update to its translation services that, thanks to new machine learning techniques, promises significantly improved translations between a large number of language pairs. Based on its Project Z-Code, which uses a "spare Mixture of Experts" approach, these new models now often score between 3% and 15% better than the company's previous models during blind evaluations. Z-Code is part of Microsoft's wider XYZ-Code initiative that looks at combining models for text, vision and audio across multiple languages to create more powerful and helpful AI systems. At its core, the system basically breaks down tasks into multiple subtasks and then delegates them to smaller, more specialized models called "experts." The model then decides which task to delegate to which expert, based on its own predictions.

Microsoft Translator now works across 103 languages


Google Translate might be the go-to translator service for most people, but Microsoft's Translator is catching up with the addition of 12 new languages and dialects. Microsoft Translate now supports 103 languages with the addition of 12 languages spoken by 84.6 million people: those languages include Bashkir, Dhivehi, Georgian, Kyrgyz, Macedonian, Mongolian (Cyrillic), Mongolian (Traditional), Tatar, Tibetan, Turkmen, Uyghur, and Uzbek (Latin). Google announced support for 108 languages in Google Translate after a rare update to language support last February, which added Kinyarwanda, Odia, Tatar, Turkmen, and Uyghur to the list. SEE: BYOD security warning: You can't do everything securely with just personal devices Both companies are using artificial intelligence in their cloud infrastructure to reach different language groups across the world. "With this release, the Translator service can translate text and documents to and from languages natively spoken by 5.66 billion people worldwide," the Microsoft Research group said in a blogpost.

Criminals use fake AI voice to swindle UAE bank out of $35m


In brief Authorities in the United Arab Emirates have requested the US Department of Justice's help in probing a case involving a bank manager who was swindled into transferring $35m to criminals by someone using a fake AI-generated voice. The employee received a call to move the company-owned funds by someone purporting to be a director from the business. He also previously saw emails that showed the company was planning to use the money for an acquisition, and had hired a lawyer to coordinate the process. When the sham director instructed him to transfer the money, he did so thinking it was a legitimate request. But it was all a scam, according to US court documents reported by Forbes.

Microsoft improves Translator and Azure AI services with new AI 'Z-code' models


Microsoft is updating its Translator and other Azure AI services with a set of AI models called Z-code, officials announced on March 22. These updates will improve the quality of machine translations, as well as help these services support more than just the most common languages that have less available training data. The new Z-code models use a sparse "Mixture of Experts" approach, which Microsoft execs described as being more efficient to run because it only needs to engage a portion of the model to complete a task. The result, according to Microsoft: Massive scale in the number of model parameters while keeping the amount of compute constant. Microsoft recently deployed Z-code models to improve common language-understanding tasks like name entity recognition, text summarization, custom text classification and key phrase extraction across its various Azure AI services.

From search to translation, AI research is improving Microsoft products


Until recently, a multinational company looking to help customers around the world book international travel would have had to build separate chatbots from scratch to converse in French, Hindi, Japanese or other languages. But thanks to artificial intelligence research breakthroughs that have enabled algorithms to more accurately parse nuances in the way different languages express concepts or structure sentences, it is now possible to build a single bot and use Microsoft Translator to translate questions and answers accurately enough for use in multiple countries. Over the past few years, Microsoft deep learning researchers were the first to achieve human parity milestones in developing algorithms that could perform about as well as a person on research benchmarks testing conversational speech recognition, reading comprehension, translation of news articles and other challenging language understanding tasks. Now, the benefits of those AI research breakthroughs are making their way into products from Azure to Bing. Search engineers are borrowing lessons from Microsoft AI researchers who developed a new deep neural network model that can learn from multiple natural language understanding tasks at once.