What kind of experience is working at Microsoft Asia Research?

Lead: Microsoft Research Asia was founded in 1998 in Beijing, China. It is Microsoft's largest basic research institution outside the United States. At present, Microsoft Research Asia has 20 research and engineering teams, mainly engaged in natural user interfaces, next-generation multimedia, digital-centric computing, Internet search and online advertising, and computer science based on the five major areas of research.

"People" published a cover article entitled "The Whampoa Military Academy of the Chinese Internet" on July 18th, 2016, which caused a heated debate. The beginning reads: Microsoft Asia Research Institute is the president of Baidu, the father of Alibaba Cloud, Kingsoft CEO and Xiaomi. The cradle of the president, more than 500 people who are active in the IT industry in China, and more than 100 people are teaching at the top universities in China.

In this issue, we specifically invited Dr. Zhang Fuxin, an associate researcher at Microsoft Research Asia, to talk about his experience working at Microsoft Research Asia and the Microsoft Research Asia in his eyes.

Zhang Fuzheng, Ph.D. in Computer Science, University of Science and Technology of China, and a bachelor's degree in computer science and financial statistics. He had an exchange visit at the University of Queensland, Australia. He is currently an associate researcher at Microsoft Research Asia. His research interests include user models, recommendation systems, deep learning, sentiment detection, social networking, spatio-temporal data mining, pervasive computing, and large-scale systems. He has published in major conferences and journals in the field. A number of articles, such as KDD, WWW, Ubicomp, TIST, etc., have won the ICDM2013 Best Paper Award, and have served as invited reviewers and executive committee members of the AI â€‹â€‹field conference for many times.

Dr. Zhang first briefly introduced the experience of coming to Microsoft Asia Research Institute: The first time was in the next semester of senior year. At that time, he had an internship for six months and completed the graduation design here. As a joint doctoral student at the China University of Science and Technology and Microsoft Asia Research Institute, Dr. Zhang completed all courses at the school for the first year, and then spent the next four years of doctoral career at Microsoft Research Asia. After graduating from the doctoral degree last year, Dr. Zhang joined the social computing group led by Dr. Xie Xiu of the Institute.

Today, Dr. Zhang shared the work and life here with people from the perspective of a research institute young person.

Before you come in

What unique culture does Microsoft Research Asia have made it known as the "Huangpu Military Academy" in the AI â€‹â€‹world?

This is a big problem. I mainly answer this question from three aspects.

As a crossroad between academia and industry, Microsoft Asia Research Institute has played such a role: On the one hand, he has established a deep connection with colleges and universities, and he has collected from colleges and universities every year those who have very solid academic abilities. At the beginning of the establishment of Microsoft Asia Research Institute, an academic cooperation department was established. This department is responsible for the cooperation with universities and the cooperation with universities to cultivate talents. For example, during my doctoral period, I can benefit from long-term internships at Microsoft Asia Research Institute. The joint research project of the institute and the university. On the other hand, as a high-tech company that stands at the forefront of the industry, Microsoft is not only able to do in-depth research at the Institute, but also often embraces the fear of science and technology to implement technology products. Our research results are also often applied. In various black technologies released by Microsoft. Our researchers and the product group have close cooperation, where you can fully exert your strengths, participate in it, and explore the future of technology.

The Microsoft Research Institute in Asia was established in 1998. The profound technology deposition in the past 20 years has given these recruiting talents a broad platform. At the same time, the Academy encourages everyone to conduct free research under the drive of interest and enable them to play in various fields. Their innovative ideas. The position of our institute is to allow employees to do research and innovation with the utmost care. The Institute's cultural atmosphere is to praise and encourage people who have a strong interest in academics and have an utmost pursuit of science and technology. Working here, we don't have deadlines (except, of course, you decide to send a paper, but no one tells you to send it). Because if one thing you can plan to do today, it is not innovation, everyone can do it. Research is to make researchers think about problems in a very free environment, but in the process of searching for good ideas must have experienced many failures, and explore the direction of success in trial and error. Here researchers have full autonomy. In research, especially young scientists who want to innovate, they must be free.

Many graduates are gathered in the institute and many of them are world-class experts in their fields. For example, Dr. Hong Xiaowen, our dean, is the top expert in the field of speech recognition. Guo Baining, the vice president, is one of the few domestic ACM Fellow. In addition to leadership, the cattle people around us are also everywhere. For example, Song Ruihua, a researcher from our group, is not only the top scholar in the college entrance examination, but also a big player in the field of information retrieval. In such an environment, young people not only can quickly learn the most advanced knowledge in their field, but also can see the latest developments in other fields, which is conducive to fostering profound insight and a broad perspective.

How does it maintain its forward-looking and farsightedness in work?

Let me talk about the three missions that are often mentioned in the Institute:

The first is "advance the state of the art of computing". It means that we must do the most cutting-edge research. For example, we studied artificial intelligence long ago and made a tangible contribution to the academic community in this field.

The second is rapid transfer of innovative technologies into Microsoft products. We must put the latest research results into the actual products.

The third is next big thing - incubate for the future. We not only focus on the present, but also look to the future to do some seemingly challenging, even completely unobtrusive research, such as quantum computers and HoloLens and other black technologies.

Our dean often said that Microsoft Research is making money for Microsoft: If the sales department is responsible for selling Microsoft products and making today's money, the product development department is responsible for designing and developing Microsoft's products for 5 years, earning tomorrow's money; Then the Institute needs to look further, to ensure that Microsoft will have the technical lead in the next 5 to 10 years, that is, responsible for making money.

Therefore, whether it is from the mission or the positioning of the research institute, we always ensure that we remain technologically advanced and forward-looking.

In fact, in our daily work, we also deeply participate in the top conferences in the research field, transparently publish the latest research results, communicate with the most talented people in the industry, and always stand at the forefront of technology. In a free and relaxed atmosphere, the researchers here are guided by interest and carry out some forward-looking projects. As far as I am concerned, Iâ€™m currently doing some forward-looking projects related to emotional computing. I hope this project can not only accurately identify the userâ€™s emotions, but also enable chat bots such as Microsoftâ€™s Xiao Bing and users to build true emotional relationships. Depends on and resonates, which is a difficult task for these robots.

The first time in the senior year is in the next semester.

Can you combine some of the details of work and life and talk about what kind of experience will be after Microsoft Asia Research Institute?

I talk about the feelings of two stages of internship and work.

My Ph.D. career was basically spent at the Microsoft Research Institute in Asia. Now, in retrospect, this is a very fortunate event. Internship here is not the same as school and company. The relaxed and free academic atmosphere here allows me to explore the research I am interested in. There are also cutting-edge scientific and technological advances that allow me to observe the trends of technology. Most importantly, here is a place where both IQ and emotional density are extremely high. When I encounter a problem, I can always find someone to help me expand my thinking. Many of the ideas I had in my research work were discussed together with the Institute's little friends in the process of eating, drinking, and drinking.

Not only does the Academy have a lot of world-class scientists who can learn to worship, but the Academy has maintained close ties with outside academic masters or industry leaders. It often invites various cattle people to talk to them, basically every week. The rhythm of at least 2 to 3 lectures, so here is a very good platform for expanding horizons.

In addition, the Institute also strongly encourages students to participate in various academic activities, such as sponsoring students to attend high-quality international academic conferences, and often organizing activities such as outstanding doctoral student forums and student technology demonstration festivals to allow students to demonstrate and communicate. His own recent research results.

Another feature of the institute is its diversity. Not only do we have small partners from other Asian countries, but also often come from Europe and the United States as researchers and interns. We can experience a variety of cultural influences and styles of doing things from a foreign country without going abroad. Is very conducive to the cultivation of international horizons.

In addition to the work, the life of the intern on this side is also colorful. There is a great variety of coffee to play with you. For example, there are table football equipments in many floors of the company. Many of my classmates used to learn a few things in their spare time, but in groups of three or more, everyone was relaxed. Not only did the skills improve, but they also met many like-minded individuals. partner. Every week, everybody will also organize a variety of board games and sports activities. Football, basketball, badminton, and people who are interested can participate freely. Our slogan is work hard, play harder. Life and work are often connected, and people who have a passion for life tend to work very well. There is also an event called family day every year. Researchers and interns often bring their families together to participate. There is a big family-like feeling. This is a picture of our group participating in the family day this year.

Here comes a soft advertisement. This video introduces the intern's work and life here. Everyone is interested to have a look (video link).

After becoming a full-time employee, there will be some different feelings besides being able to enjoy the various benefits before. The first is the positioning of oneself. During the internship, it is more important to complete the company's project from the perspective of participation. Now it is necessary to assume the role of the project leader. Not only do we need to obtain relevant resources, but also need to determine the most appropriate solution. The progress of the plan and the coordination of the project is not only the executor of the project but also the manager of the project. Now I have taken some students on the ground. I feel deeply responsible in this process. You only need to do your job well when you are practicing. Now you must not only do your own job, but also often think about how to allow students to grow rapidly here, so that students can not only harvest research results, but also to foster independence. Research and innovation capabilities.

You want to be a cow person and get close to the cow person first

It is said that Microsoft Asia Research Institute has two titles in the industry. One is called the Ph.D. organization in Asia and the other is PhD killer. What is the difference between a researcher at Microsoft Asia Research Institute and a doctoral student at school?

In fact, I basically read Dr. here. Let me talk about why you choose to read a doctorate here.

As mentioned earlier, Microsoft Asia Research Institute is a crossroad between universities and industry. I can not only get access to the most advanced scientific research, but also get insight into the latest developments in the industry. My personal preference is to apply the latest research results to actual products, and Microsoft Asia Research Institute is the ideal platform to meet this condition. Here, our research inspiration often comes from the product development process. When we face practical problems and there is not a good solution yet, this issue is often a good scientific research issue. We need to adopt innovative ideas. explore.

The Institute also has a wealth of resources. There are many big cows and even world-class scientists and very powerful students from all corners of the world. To become a cowman, you should go to a place that is closest to the cows. In addition, there are powerful hardware devices and data resources here. In the era of big data and artificial intelligence, the massive user data in the company allows us to do a lot of interesting problems, and a large number of GPU and other computing resources can make our research change. It's very efficient.

In your research results, are there examples that have been applied to the product to share with us?

Let me talk about the lifespec project. With the widespread collection of user data in recent years, the user portrait area is very popular and many companies are keen to more accurately target their target customers. Lifespec is such a project that collects user data exposed by each platform and correlates it. The model can graphically describe the characteristics of users and groups.

For example, these two diagrams show the different life styles of financial practitioners and software practitioners. Those engaged in finance prefer to look at financial books or go to the bar to sign in, and those who are engaged in software prefer to read books on computers and programming. Check in at the office.

This is the homepage of the lifespec project. Everyone has time to look at (project homepage link).

At present, this project has cooperation with many products. One of them is the image of college graduates who have gone online this summer. Users can input school-related keywords in Microsoftâ€™s search engine Bing, and they can see the characteristics of the school graduate in the results, such as the most enthusiastic work units. Frequently visited locations, etc. In addition, we are still collaborating with Microsoft Cloud to provide user portrait services on the cloud. Many companies expect that this service on the Microsoft Cloud can better screen their target customers for more accurate advertising.

What are your preferred research directions or technology hotspots in the future? why?

Because I have been doing research in the field of artificial intelligence, and I am familiar with this field, I mainly talked about some of my personal preference.

Personalized robots will play an increasingly important role in the future. There are a variety of robots on the market today, including physical smart devices such as Amazon's Echo, social networking home robot Jibo, and virtual robots on the Internet such as Microsoft's Xiao Bing and Xiao Na. These robots have begun to enter all aspects of life, such as chatting with you, scheduling your schedule, controlling your home's electrical equipment, telling stories to children, etc., and providing more information and services for us with more and more natural personification. And with the accumulation of user data, these robots will become more and more personalized, just like your friends or family members. For example, Echo can give family members a gossip about what it collects from friends around you at dinner.

Emotional computing is also a promising direction in the future. There is a movie named "Her". After the actor has finished a heartbreaking love run, he falls in love with the computer operating system. A "girl" named "Samantha", this movie is about The emotional communication between machines is vivid and amazing. In fact, in addition to emphasizing the completion of tasks and functional enhancements, people generally hope that they can own and use more humane computers and smart devices, and even begin to look forward to them being able to have emotions like human beings, and being able to naturally and harmoniously interact with humans. Can understand the joys, sorrows and pleasures of the people and see the machine acting.

In fact, Minsky, one of the founders of artificial intelligence, clearly pointed out in the 1980s that â€œthe question is not whether smart machines can have emotions, but whether machines without emotions can realize intelligenceâ€ and the importance of emotional computing for artificial intelligence can be seen.

In fact, we are also working closely with Microsoft Xiao Bing to help Xiao Bing and users build evolving emotional communication based on the emotional signals reflected in user data, such as lonely words, happy expressions, and anger-specific tones.

The fire of a comet can be a

With your background and ability, whether you are involved in academia or industry, whether you stay in school or enter the company, you will certainly have a lot of choices. Why did you choose Microsoft Asia Research?

As mentioned earlier, I personally like to do the most cutting-edge research, and I also like to apply these research results to actual products. Microsoft Asia Research Institute, as a bridge between the academic community and industry, is an ideal platform to meet this condition.

However, we also noticed that in the end many people â€œrun awayâ€ from Microsoft Research Asia. Many of them entered the industry or simply started their own business. How do you see this?

I personally feel that I must look at this issue with a sense of commonality. The flow of talents in companies, especially Internet companies, is a normal phenomenon. A certain amount of talent flow will instead make companies stable but not rigid and live without confusion.

As Dr. Shen Xiangyang, our former dean and current Microsoft executive vice president, said: "Microsoft Research Asia cannot always be the only successful computer organization in China."

Because with the development of the times, despite the fact that we have very great advantages under any circumstances, the gap between other companies and us will continue to shrink. This is only a matter of time and in what form. Maybe it's normal for us to train more talents to go to local companies, go to local universities and other universities, or go to startups now.

In fact, for us, only one thing is important:

Microsoft Asia Research Institute can continue to innovate and continue to do the most amazing things.

If you want to do scientific research or want to have an impact on scientific research, Microsoft Asia Research Institute is the best institution in China and even Asia-Pacific region.

Actually, there is a saying that is very good for those who leave the institute. "The gathering is a fire, scattered as a starry sky." Although they did not continue to work in the Institute, many people are still closely related to the Institute. Contact and cooperation. Citing the words of our director, Dr. Hong Xiaowen: â€œFrom the perspective of the institute, the position of the institute is very clear. This orientation will not be suitable for everyone, nor will it suit everyone forever. Some people think that I want to If we do not concentrate on doing research and innovation here, we will be able to stay here. Some people say that I now have other ideas. I want to achieve it, whether it is products, whether through schools or so-called entrepreneurial ventures. We will all wish them success, so even if they go out, we still maintain a very good relationship with these people. We are still Partner."

Can you combine your own personal experience and talk about what Microsoft Research Asia is looking for when selecting talents?

In fact, Dr. Shen Xiangyang summed up: Microsoft Asia Research Institute recruits people to "three good" - the first mathematics is good, the second programming is good, the third attitude is good.

About mathematics is good, this is the necessary skill to do scientific research. When I interviewed, I wrote a lot of mathematical derivation in machine learning on the small blackboard. Regarding programming, I think this is a necessary skill for computer students. No matter how good the model is, donkeys are horses. With regard to good attitudes, today's scientific research is very "unfortunate" because the chance of getting a pen and a piece of paper immediately becomes less and less. Various software and various algorithms are not necessarily written by one person, but are discussed by a team. Therefore, you are willing to cooperate with others, and it is important to feel that you can learn from others.

Dean Hung also often told us that it's very important to do scientific research because we have passion (passion, enthusiasm), because scientific research is very difficult at times, and you have to create a road that others have never traveled before. If you don't have enough passivity to support you, it's really hard to stick with it. So enthusiasm for scientific research is also an important aspect of selecting talents here.

In general, the Institute likes to have a solid academic and technical foundation, good communication skills and mentality, and smart people.

Finally, what advice would you give to those Ph.D and MS graduates who would like to enter Microsoft Research Asia in the future?

Because scientific research is an important part of the work in research institutes, it is very important to have good research background and capabilities, such as publishing articles, participating in academic events and conferences in well-known international conferences or journals.

The ability of programming is also very important. We often deeply participate in the research and development of products, so we need good programming skills to implement the algorithm model into the actual product. If you can have experience with actual projects or company internships, you can prove that you have had experience in actual systems before. This is also a very good addition. For example, students in the field of data mining can participate in some contests on Kaggle. If they can achieve good rankings, this is an intuitive manifestation of their algorithm model and programming capabilities.

Last but not least, there is a strong interest and enthusiasm for research in technology. Because the process of scientific research is often arduous, it is often difficult to achieve good results on this side if it is only done as a job rather than driven by a strong interest.