close
AI

Leave MSRA and join AI Unicorn: After the scientist leaves the ivory tower

b12b5c8e-c4dd-11e9-98aa-aaaa00151f43.jpg

Editor’s Note: This article comes from WeChat public account “CV Intelligence” (ID: CVAI2019), author Yu Yangyang, editor Zhang Lijuan, 36氪 authorized to publish.People who understand business may not understand the most complex and profound technology, but scientists who have worked in the academic field for decades have never been lacking in the industry after they have stepped out of the laboratory’s one-acre three-point land.The opportunity to do it.“Beijing’s schools have concentrated almost the best talents of China’s more than one billion people.” As Tang Xiaoou, founder of Shangtang Technology and former director of the MSRA Visual Computing Group, said, Microsoft Research Asia (MSRA), located in the hinterland of Zhongguancun Science Park,It is connected to Tsinghua University and Peking University in the north, the People’s Congress in the south, the North Polytechnic in the south, the North China Airlines and the North Post in the east. Since the near-water building has the first month, it has gathered a group of the best scientific and technological talents in China.Tang Xiaoou, Wang Xiaogang, Lin Dahua, Yang Fan of Shang Tang, Sun Jian and Yin Qi, who are despised, Zhu Xi, Lin Chenxi, Yan Shuicheng of Yitu, Huang Dinglong and Matt Scott of Codelong Technology, Chinese AI pioneers and entrepreneurs are here.When they met, they were colleagues and apprentices who worked hard at MSRA.Out of the MSRA, into the fiery AI entrepreneurial wave, they turned into the company’s founder, CEO, chief scientist.MSRA provides them with an excellent platform to do research, communicate and get to know each other.Even if we leave the institute and move to the commercial sea, we are faced with another world. The identity has changed and the role has changed. What remains unchanged is that scientists have for many years adhered to the trust and friendship that was forged in the front line of scientific research.In contrast, Zhongguancun is like a flowing feast of Hemingway’s pen. “If you are lucky enough to live in Paris when you were young, then you will be with you wherever you are in the life.” Surrounded by the familiar MSRA, young AIThe unicorns grew up here.Before they could have their own products, these emerging technology companies often required the endorsement of several well-known experts, so they had the existence of “chief scientist” and “dean of the institute”.Why is AI company keen to set up a research institute and invite academic scholars to sit down? Zhang Wei, a professor of computer science and mathematics at the Hong Kong University of Science and Technology and former director of Tencent AI Lab, said to CV Intelligence that “the institute has several functions, one function.It will support some products on short-term projects; the other is that it will also do some technical reserves to accumulate some products; there is another point that I think is important but neglected, the institute as a company talentReserves, these talents can produce value when the industry changes rapidly.” In Zhang Wei’s view, Microsoft’s successful transformation in the wave of multiple rounds of the times has returned to the world’s first market capitalization, which is closely related to its talent pool.”For example, Microsoft,” Zhang told CV intellectuals. “It turned out to be a software company, but when it was doing search, although it did not achieve the level of Google, it could get up quickly, including doing cloud computing now.Why can it be one of the companies with the highest market value? Because of the reserve of technology and the accumulation of talents, there is no way to do these things without the reserve of talents in the transformation of the company.” “Chief Scientist”, “Research Institute”The long-term leaders led their respective companies, constantly breaking through technical problems, and sharing a lot of pressure for the founders and CEOs.Wang Xiaogang and Shang Tang’s Hong Kong Chinese “Army” Wang Xiaogang is one of the co-founders of Shang Tang and the dean of Shangtang Research Institute.When Wang Xiaogang and Tang Xiaoou were young, they experienced the highlights of many academic careers.Both graduated from the Chinese University of Science and Technology and both went to the Massachusetts Institute of Technology (MIT) to study.When Wang Xiaogang went to the Chinese Multimedia Lab in Hong Kong, he also studied under the law of Tang Xiaoou, and he was the proud student of the latter.With the establishment of Shang Tang, Wang Xiaogang and Tang Xiaoou joined the road to the north and served as the dean of Shangtang Research Institute.At that time, Zhang Hongjiang, then vice president of MSRA, sent an invitation to Tang Xiaoou, who is a teacher at the Information Engineering Department of the Chinese University of Hong Kong. He hoped that he would serve as the director of the media computing group.Although reluctant to leave his wife and children in Hong Kong, Tang Xiaoou still joined the computer vision group under the new MSRA Dean Shen Xiangyang, and the media computing group was also incorporated into the computer vision group, which was taken care of by Tang Xiaoou.”I promised the next day. Harry (Shen Xiangyang) is also afraid of long night dreams. The next day we have finalized many details. Without any interview, I became a researcher within a few days. SaturdayI bought a house. That week, I felt like I was in a relationship with two people who were afraid of each other’s remorse and hurriedly got a marriage certificate.” Tang Xiaoou once recalled the experience in a self-report, although he was in a hurry, he did not feel regret at all.”I have been admiring the Institute for a long time. The Institute is very much like Mingming in my heart. It has the momentum of the world’s first name.” As Tang Xiaoou said in his self-report, within three years of MSRA, Tang XiaoouLeading the team to publish more than 60 papers in first-class computer vision conferences (ICCV [2], CVPR [3], ECCV [4]).Another advantage of sending these papers is that they attract a lot of good students. “I have seen many very good students in these years. Some of them have not been described as excellent. They can only be said to be geniuses.” Wang Xiaogang is exactly the same as Tang Xiaoou at MSRA.”The first genius student.”In the eyes of Tang Xiaoou, Wang Xiaogang published five CVPR/ICCVs at the master’s level.In addition, Tang Xiaoou also quite appreciates Wang Xiaogang’s character. “His talent and character are so outstanding that I did not hesitate to marry my sister.” Wang Xiaogang, Dean of Shangtang Research Institute, from 2009After graduating from MIT, Wang Xiaogang returned to Hong Kong to teach in China and worked side by side with Tang Xiaoou.Wang Xiaogang is not expected by Tang Xiaoou. In the 2014 ImageNet large-scale object detection mission competition, Wang Xiaogang, Ouyang Wanli and Tang Xiaoou led their team DeepID-Net to take the second place with a 40.7% record.ImageNet is the annual “Olympic” in the field of visual identity. It has a profound impact on computer deep learning. The technological advances made on ImageNet often have an important impact on other computer vision problems.A few months later, the DeepID-Net team increased this score to 50.3%, surpassing the long-term Google as the number one, achieving the highest detection rate in the world.This achievement was published at the 2015 International Conference on Computer Vision and Pattern Recognition (CVPR), the highest record of the ImageNet challenge in the world at the time.Prior to this, the DeepID team led by Tang Xiaoou and Wang Xiaogang also defeated Facebook in the LFW face recognition challenge, and for the first time in the world, the face recognition algorithm surpassed the accuracy of human eye recognition.DeepID-Net team, the right fifth is Tang Xiaoou, the right third is Wang Xiaogang in 2014. With the establishment of Shang Tang, Wang Xiaogang also opened the life to and from Hong Kong and Beijing. He served as dean of Shangtang Research Institute.The world has witnessed the numerous problems and breakthroughs in the process of landing the algorithm. It has also experienced the whole process of gradual expansion and gradual contraction of the team from the initial establishment of the Shangtang Research Institute from scratch.The research institute at the peak has reached nearly 2,000 people.Not only Wang Xiaogang and several other co-creators, CV intellectuals have communicated with many insiders of the business soup, many doctors of the Hong Kong Chinese Multimedia Lab (mmlab) will choose to join Shang Tang after graduation, and there will be teachers and students.A ties, the doctor of Hong Kong Chinese is naturally more trusted by the company.After leaving MSRA, Sun Jian took the helm of the Institute. In July 2016, Sun Jian, the former researcher of Microsoft Research Asia, joined the contempt and caused heated discussion in the industry.In fact, Sun Jian was a teacher at the internship of Microsoft during the undergraduate period.Sun Jian has been working for MSRA for more than ten years and has been a Principal Investigator since he was a soldier.In recent years, Microsoft’s Asian research has been affected by various kinds of technology startups. More and more senior researchers have gone to MSRA and become co-founders and CTOs of startup companies.Some of the earlier MSRA people such as Kai-Fu Lee founded the investment institution innovation workshop and served as the chairman, Lin Bin served as the president of Xiaomi, Zhang Yaqin as the president of Baidu, and Zhang Hongjiang once served as the CEO of Jinshan Software.In this context, Sun Jian chose to leave and join the contempt.”I am a Tsinghua undergraduate from 2006 to 10 years, and at the same time I have been internship for more than three years at Microsoft Research Institute (MSRA). That piece may be a place for my more specialized research start-up,” InchI remembered the time at MSRA when I was interviewed by the Innovation Workshop.At that time, it was the first time that Inge and Face Recognition met intimately.And then, inside the contempt, there is also a research institute responsible for basic research and development.This is regarded as one of the core competencies of its technology.The arrival of Sun Jian adds to the glory and strength of the Institute.On the reputation, Sun Jian is the chief researcher of MSRA; on the strength, Sun Jian has achieved countless achievements in the field of academic research; on the human condition, Sun Jian is the guide for the start of India’s scientific research, three years of co-existence is enoughThe two establish trust and friendship for each other.In 2001, Sun Jian, a joint researcher of Microsoft Research Asia and Xi’an Jiaotong University, chose to stay at Microsoft Research Asia and put his research direction on computational photography, face recognition and deep learning-based image understanding.In 2004, Sun Jian participated in a graphic project research for the first time as a Ph.D., and soon became a researcher at Microsoft Research Asia.Young, but academically successful, in 2010, with the strong influence generated in the academic world, Sun Jian was selected as the “outstanding young innovator of the world under 35 years old” by the MIT Technology Review.In 2015, the team led by Sun Jian won the championship of all three major projects of image classification, object location and object detection by using the 152-layer neural network in the ImageNet Computer Recognition Challenge.From the outside world, Sun Jian seems to have a treasure that can’t be dug. For example, he has participated in the development of a famous “How-Old.Net”, which can be based on user-uploaded photos.In order to quickly locate, detect, identify and judge the age, Sun Jian also has more than forty international or US patents. His and his team’s research results have been widely used in Microsoft Windows, Office, Azure,Bing, Xbox and other products. Inch said that from the talent citation, Sun Jian is his mentor in Microsoft for several years, and is also the best computer vision expert in the Chinese circle. More importantly, he and Sun Jian are each other.Recognized, the field of artificial intelligence is also the future trend. Say goodbye to 360, Yan Shuicheng joins Yitu and the stars surrounded by Shang Tang and contempt, and Yi Tuo, founded by Zhu Xi and his classmate Lin Chenxi, is located in Shanghai and acts.Low-key. Since the company was founded seven years ago, there has been almost no news that the industry’s big bulls have joined the map. According to the map, there are no industry veterans like Tang Xiaoou and Yao Zhizhi who are sitting behind the town.In July, Yan Shuicheng, the vice president of the original 360 Group and the dean of the Artificial Intelligence Research Institute, officially announced his participation in Etu as the chief technology officer of Yitu. Yan Shuicheng went to school and spent 9 years studying mathematics and applied mathematics at Peking University School of Mathematical Sciences.Really connected with artificial intelligence, it entered Microsoft Research Asia in 2001. During MSRA, Yan Shuicheng from the face recognition and intelligent video surveillance expert Li Ziqing and the originator of computer video retrieval, the founder of Microsoft Research AsiaA Hongjiang. In MSRA and Zhu Xi and Lin Chenxi met, eventually became the opportunity for Yan Shuicheng to join the map. “The early CV startups mostly have a relationship with Microsoft or Professor Huang Taotao, I have stayed in two places, soEveryone is very familiar with each other.I have been in the same office of MSRA for a long time with the founder of Yitu, and I have had an intersection with the co-founder Lin Chenxi at MSRA. At the same time, I also know the development process of Yitu. Therefore, when Zhu Xi invited, he joined the map.The decision is also very simple and quick.In 2006, Yan Shuicheng left MSRA and went to the University of Illinois at Urbana-Champaign (UIUC) to study abroad. He studied under the American Academy of Engineering and computer vision master Thomas Huang. In 2008, Yan Shuicheng joined the National University of Singapore and gradually became a machine learning student of NUS.Calculate visual leaders and become tenured at the National University of Singapore.Before joining Etu, the experience of Yan Shuicheng 360 Group also brought important industrial practical experience to him with profound academic background.In 2015, Yan Shuicheng joined the 360 ​​Group.During the 360 ​​Group, he led the team to many awards in the field of computer vision competition. More importantly, under his leadership, the Institute’s artificial intelligence technology has been widely used in 360 full range of products, including live broadcast, intelligent hardware., search and information flow and other services.“Let AI land” is also one of Yan Shuicheng’s work philosophy. He has emphasized the importance of AI technology in the industry in many public occasions.The addition of Yan Shuicheng undoubtedly added strong technical strength to the company, and the position of chief technology officer for the post-no industry veteran and former chief scientist.As Chief Technology Officer, Yan Shuicheng will work with the Etu team to further strengthen AI basic research.At the same time, finding more technical talents according to the map is also one of the important tasks of Yan Shuicheng.Previously, an insider told the CV intellectuals that since August last year, the team of Yitu has expanded rapidly from the previous three or four hundred people in 1989. “It is obvious that this year the companyAt the beginning of the effort, all business lines are ready.” The arrival of Yan Shuicheng will also make Etu more powerful in the process of expansion, and more AI talents will come to the fore.Leaving the ivory tower in 2008, Huang Dinglong, who entered MSRA, met with Matt Scott, then senior director of MSRA research and development, and the two became attached.For many years, Matt has been the top 1% of Microsoft’s top performers and managers, and has led the development of the “Bing Dictionary.”Huang Dinglong is a product manager at MSRA.The combination of Matt and Huang Dinglong was derived from the development of the Bing Dictionary, which was called “Golden Partner” in the hospital.Huang Dinglong is optimistic about the application of AI in the retail field. “In the future, all supermarkets and shops may be able to operate 7*24 hours. Customers can push the cart and walk through the camera to know what they bought and pay.How much money, the trouble of queuing and checking out will be solved by technology. Put a camera in the store, this camera will work like a cashier who will never rest.” The application prospect of product identification is not as clear as today.Huang Dinglong realized the potential of this field.After Matt and his Chinese girlfriend married in 2014, Matt and Huang Dinglong unanimously optimistic about the AI ​​revolution that deep learning will bring, and the two will meet each other.Shortly thereafter, Codelong Technology was founded, taking each word in the name of two people.MSRA seems to be a besieged city. During the past 20 years, people have been rushing in, and people have left with their dreams.It is not only the privilege of MSRA people that technology people go to sea to start businesses and stand on their own feet. Scientists follow them and help each other.At the beginning of the business, Codelong Technology, the middle blue clothes for Huang Dinglong, the black for Matt Scott “people in this industry go out to stand on their own,” once an AI industry insiders sighed to CV intellectuals, “MSRA has a wave, BaiduThe wave of Hong Kong Chinese and Hong Kong technology, the Chinese Academy of Sciences also has a wave. With the AI ​​into the undercurrent commercialization stage of competition, more scientists began to leave the ivory tower, leading the research department of the technology company.Ma Yuanyuan was the person in charge of the defensive security business line. In 2018, he left to despise and founded his own AI company.In March of this year, Zhai Si recruited Chinese AI scientist Shen Shengmei to join Yu Si as the chief scientist, and at the same time set up the Singapore Research Institute, responsible for the team building of the institute and the exploration of artificial intelligence frontier technology.Interestingly, Ma Yuan and Shen Shengmei were “sisters and sisters” and both graduated from Xi’an University of Electronic Science and Technology.Shen Xuemei graduated from the Northwest Institute of Telecommunications Engineering (now Xi’an University of Electronic Science and Technology) in 1984. He is a professional graduate of the first batch of intelligent science and technology at Xi’an University of Electronic Science and Technology.There is no such thing.Jiang An spent nearly a decade at the Institute of Automation of the Chinese Academy of Sciences.In 2012, he left the Chinese Academy of Sciences and entered Nokia at that time. However, it didn’t last long. In 2013, Microsoft acquired Nokia, followed by Microsoft’s layoffs in 2014. The ruling colleagues later went to IBM and Intel.For Jiang An, this layoff storm has just triggered his entrepreneurial thinking. “I always work for the capitalists, but I don’t want to do it myself. So in 2014, when I was laying off employees, I took a brother to start a business.Let’s make a living!” When it was said, Jiang An founded Zhongke Huiyan with several doctors and engineers from the Chinese Academy of Sciences, Nokia, Microsoft, Samsung, Siemens, Huawei, and Toyota. The company’s chief scientist is an associate researcher at the Chinese Academy of Sciences.The director is a researcher and doctoral supervisor of the Chinese Academy of Sciences. The entire research team is led by his most trusted Chinese Academy of Sciences teachers and brothers.One end connects academics and one ends connects businesses from academic circles to industry.Under the wave of AI commercialization, scientists are constantly coming out of research institutions and rushing into the entrepreneurial craze.Because they know technology and are famous, scientists are often the best “posts” of young AI companies.With the fame and achievements accumulated in the academic circle for many years, they recruited talents for the company and formed a technical army that can “winter” and “hit”.Zhang Wei told CV that the most important significance of the research institute or AI Lab is to accumulate the long-term development of the company. It reflects the forward-looking technology of a technology company.“If you want to solve the problems on your existing products right away, it must be the best and fastest in the product department,” Zhang Wei told CV Intelligence, but at the Institute “there are some more basic ones that are not necessarily right away.It can be used, maybe one year, three years, five years. “Scientists are “posts”, but they are more than just “stage pillars.”People who understand business can not understand the most complicated and profound technology, but scientists who have worked in the academic field for decades have never been lacking in the industry after they have stepped out of the laboratory.The opportunity to show your skills.Today, the demand for commercialization of AI has become so urgent. They are cautious about the balance between algorithms and landing, and they are anxious about the anxiety of the founders.References: The world’s first Ming – Professor Tang Xiaoou’s self-reported story.Tang Xiaoou.Despise technology and India: The vent is waiting, not coming.Innovate China’s first season.Interview with Face++ CEO Inch: After the financing of 460 million US dollars, more super-computing centers will be built.Lei Di touched the net, Lei Jianping.Dr. Yan Shuicheng joined Etu Technology as Chief Technology Officer (CTO).According to the technology.Pilots | Code Long Technology Huang Dinglong: AI can become the master of various industries, Phoenix Satellite TV Information.Keynote speech “Foreseeing infinite possibilities”, Jiang An..