Most popular Huang Xuedong takes up the post of ch

  • Detail

Huangxuedong becomes the chief technology officer of Microsoft global artificial intelligence! Microsoft's first Chinese academician of technology is fully responsible for azure cloud ai

recently, Huang Xuedong, a 27 year veteran of Microsoft speech recognition technology and the first Chinese academician of technology of Microsoft, served as the chief technology officer of Microsoft's global artificial intelligence and was fully responsible for Microsoft's azure cloud artificial intelligence, including voice, machinetranslation, computer vision, natural language and artificial intelligence cognitive services

according to the latest appointment, huangxuedong (technology fellow, now chief technology officer of AI cognitive services) will lead the newly unified AI cognitive services engineering and research team. Huangxuedong recently led the speech and language team of cognitive organization, and is one of the world's top experts in the AI field. The voice, language and computer vision teams in the cognitive organization will move to the AI platform and report to Huang Xuedong, while the existing form recognizer, personalizer and other cognitive service engineering teams in the AI platform organization will also report to Huang Xuedong

first of all, we would like to congratulate Dr. Huang Xuedong. We expect Microsoft AI to make another success under the leadership of Dr. Huang Xuedong! Next, let me introduce Dr. huangxuedong to you

founder of Microsoft voice technology and the first Chinese technical academician of Microsoft

Microsoft is known as the "Whampoa Military Academy" in the AI field. Microsoft's global technical academician represents the highest honor for technical personnel. Up to now, only more than 20 people around the world have won this honor, including huangxuedong, the first Chinese technical academician of Microsoft and Microsoft's chief voice scientist

In 1978, at the age of 15, Huang Xuedong entered the teaching class of Hunan University (later the Department of Electronic Engineering) and became one of the youngest students on campus. He received a bachelor's degree from Hunan University, a master's degree from Tsinghua University and a doctor's degree from the University of Edinburgh

In 1993, he joined Microsoft Research Institute as a senior researcher and was responsible for establishing microsoft's speech recognition technology research team. From 2000 to 2004, as the general manager of microsoft's speech recognition technology research team, he was responsible for the research, development and market expansion of Microsoft's speech technology and products

as Microsoft's chief speech scientist, Huang Xuedong led the team to achieve a historic milestone in AI surpassing the human level:

on October 18, 2016, the Microsoft Speech team led by Huang Xuedong achieved a breakthrough in the industry standard switchboard speech recognition benchmark test, in which the word error rate (WER) of dialogue speech recognition was as low as 5.9%, For the first time, it achieved the same performance as that of professional stenographers but better than that of most people. "This is one of the breakthroughs in the field of artificial intelligence, which means that for the first time in history, a computer can recognize every word in a conversation like a human." Huangxuedong said

in 2017, the word error rate of Microsoft speech recognition system dropped to 5.1% in the task of dialogue voice transcription, and the accuracy rate exceeded that of professional stenographers. "On the surface, this progress may not be great, but in the 'last mile', every 0.1 percentage point of progress is very difficult. We must ensure that the system is free of any bugs. Moreover, from the perspective of relative error rate reduction, 5.9 to 5.1 is a relative progress of more than 10%."

in 2018, Microsoft's machinetranslation system reached a level comparable to human translation in the task of Chinese-English machinetranslation

2019, in the Stanford dialogue Q & a task, Microsoft's machine reading and understanding system set a new record, becoming the only team on the current ranking whose model score exceeded the human score

dialogue with Microsoft huangxuedong: voice and language technology is the Pearl really embedded in the crown of AI.

"only by mastering the core voice and language technology can artificial intelligence have hope, otherwise others are 'fooling'." Huangxuedong believes that speech and language are the jewels in the crown in the development of artificial intelligence

last year, Huang Xuedong made a podcast on Microsoft and talked about several milestone breakthroughs he and his team have made in voice and language technology

Huang Xuedong has always been unconvinced from imitation to self renovation. 6 the development process of the electrically controlled experimental machine explains how mastering speech recognition, translation and dialogue will make the machine evolve from "perceptual AI" to "cognitive AI", which is closer to the real human intelligence. In addition, huangxuedong also talked about how he embarked on the road of speech research, and his life course at Tsinghua University and Microsoft

host: you are the technical fellow of Microsoft voice and language group, responsible for leading Microsoft's voice work. We will discuss these in depth later, but first, as the company's chief voice scientist, please tell us what your work content is and why you do it. What motivates you to get up in the morning

huangxuedong: what we do is to ensure that we have the best voice and language technology, which can be used in various scenarios. We have a whole team doing this, because dialogue is not only the most natural way for people to communicate, but also the biggest challenge for AI. We try to really promote the breakthrough in this field, provide excellent services on our cloud service azure services, and ensure that we can satisfy customers inside and outside Microsoft who must understand the characteristics of this material. If you want to clarify the whole thing, there are three main aspects

first, we have the ability to really improve the accuracy of speech recognition, drive the naturalness of speech synthesis, and ensure the accuracy of machinetranslation, such as translating from English to Chinese, French or German. Therefore, there are many science behind this. To ensure accuracy, naturalness and delay time, these technologies are world-class. The preparation technology of fence type aluminum base lead alloy composite anode for non-ferrous metal electrowinning developed by guopiyi team of Kunming University of technology is one aspect

second, we not only provide technology, but also provide services through azure. From office to windows and Cortana, they all rely on the same cloud services. We also have edge devices, such as our voice device SDK. We want to ensure that voice technology at the edge and in the cloud is provided in a modern way, that is, cloud platform and embedded platform. This is the second aspect: the platform is modern

the third aspect is to express our love to customers. Because we have a wide range of customers all over the world, we want to satisfy consumers and ensure that the customer experience using our voice translation is first-class

these are the three key things I am actually doing: improving AI capabilities, modernizing our cloud and edge device platforms, and loving our customers

voice and language have played a decisive role in the long river of human evolution, as will the evolution of AI.

at the 2018 new Zhiyuan industry · transition technology summit, huangxuedong shared Microsoft's continuous "launching" history and technical details in voice, natural language and machinetranslation research. Huangxuedong said that speech and language have played a decisive role in the long river of human evolution, and they will play the same role in the evolution of artificial intelligence


huangxuedong said: Darwin's theory of evolution has explained that the latest scientific verification has once again proved that the similarity between human and gorilla DNA is close to 99%, and that 1% difference makes us the master of the earth. Why? Because we're smarter? Or are there other factors in the long evolutionary process? Computer vision is very important, but animals' vision is actually better than people's. night owls see better than people at night, but we are still the master of the earth. One of the most important reasons is our voice and language

2020, Xinzhiyuan will continue to explore with you

Copyright © 2011 JIN SHI