In my opinion, the goal of any doctoral degree has two points:

1, * * * to create new knowledge;

2. Cultivate world-class outstanding scholars.

When these two points conflict, 2 is more important than 1, and providing guidance and advice for doctoral students may be the most important task. In my opinion, doctoral students are the core collaborators in my research career. Without doctoral students, perhaps most of the research could not be completed. It has always been my goal to complete good research, but it is more important to train excellent researchers. I hope that the trained researchers can finally study independently without my help.

For doctoral students:

Your own research should have a certain degree of freedom and initiative;

You should have a sense of ownership of your own research;

You should be interested in the research you have done.

On the contrary, doctoral students should avoid the following situations during their doctoral studies:

Reading time is too long;

Tutors give guidance and advice, and doctoral students simply do it;

Simply think of reading blogs as a way to submit your resume.

As my student, if you think these guides are inappropriate, you can always point them out to me. I hope we can create a free, independent and meaningful reading process.

How long does it take to study for a PhD? There are probably two requirements:

Complete basic reading tasks;

You can get the job you want.

Some people can get their doctorates in three years, but I prefer to let students finish their doctorates in five to six years, especially those who want to do academic research after graduation. I suggest spending more time studying for a doctorate. There is also a question that students often ask: "What do you gain from being in charge of the church?" Let me talk about feelings:

Being a tutor is very interesting. My doctoral student is a very smart and creative person. As a tutor, I am very willing to work with them and create new things together. It is also very gratifying to see doctoral students gradually grow into completely independent researchers. In addition, I am also willing to help them get a doctorate.

Do research together. Most of my research is carried out in cooperation with students. Working with doctoral students can expose me to many interesting, impressive and difficult research problems.

How to do research?

In the postgraduate stage, the most important task is to do your own research. However, the graduate school will require the completion of some other tasks, such as reviewing papers, assisting teaching, holding lectures, setting up research groups, and operating working committees. Some of these tasks may be interesting. In the busy work, it is easy to forget the importance of your own research.

1, select a question

I hope students have a serious and positive attitude towards their research work. This is also my expectation for myself in guiding doctoral research. Please think twice before starting to solve the problem (that is, before choosing and investing time). I often think of the following questions:

If this research work is successful, what impact will it bring?

Whose life will be better because of this research?

What improvements will this study bring compared with what is being done at present?

Why is this one of the most important issues in this field?

Will it bring about major strategic changes to some extent?

Will it inspire a new system?

If you do this research, who will pay more attention?

Will government agencies pay attention?

Will platforms and industries pay attention?

Will other academic researchers pay attention to and cite this research?

Will other academic researchers pay enough attention to it and teach it to their students?

Future10,20,50? Will anyone pay attention to this research this year?

? How will this change other people's jobs?

Will others change their research work after seeing your research?

Will the work of practitioners change?

Will users adopt what you have done, discovered and created?

Will regulators adopt your findings and draft new regulations?

2. quality vs？ amount

The balance between quantity and quality is the basic pressure when doing research, especially in our research field. I call it "trade-off" because you will hear people say that you can maximize both at the same time, but in fact I have never seen a real example.

Personally, I think quality is very important of the two. If you can publish two top papers, then you can graduate and get very good job opportunities. Even if I have to choose between "only two best papers, nothing else" and "one work 14 papers", I prefer the former. In fact, in some traditional research fields, it is impossible to publish more than one paper a year.

But if we must complete the Kpi in quantity, my suggestion is that the quantity and quality are three or seven points.

This method can ensure that you can have 3-4 core papers to be proud of while completing the quantity requirements? /? Research findings/? Academic contribution. Moreover, these will become your core competitiveness in the interview. The specific operation is as follows:

Try to publish a high-quality and high-risk paper every year. During the doctoral period, there should be 3-4 works actually published. Some of these papers may not stand out in the competition, or they may not achieve ideal results in the research process (such as ineffective methods, changed background of problems, etc.). ).

? At the same time, do a simple project, that is, a paper with low return but low risk. Every two years 1.5 is more appropriate. There are many opportunities for such research to be published, but the probability of giving up halfway is very small. This kind of research may be done in cooperation with a larger research group, but you must play a more important role.

In addition, as an assistant author of at least one project. It is a research led by others, which takes you half a day or less every week. But you also provided value in the research process and won the title of assistant author. This is also a low-risk and low-return research type.

I think students who follow the following practices can graduate: 3-4 high-risk, high-return and publicly published core research; 2-4 relatively simple low-risk studies; There are also 3-4 studies as auxiliary authors.

What I think is the essential quality of doctoral students.

As a professor, I don't like students who have no ideas. My suggestion is to put forward at least one idea before the weekly meeting. The tutor had a brief discussion with the students and gave some suggestions. More importantly, we should create an atmosphere full of creativity and let innovative inspiration be everywhere.

1, where to find a good idea?

Read more papers. When I read a good paper, I will think about what to do next. What assumptions did they make? What can I do to make this paper better? Although the ideas of some papers are not suitable for me, they are just the research ideas and directions of other researchers. But some papers have a deep influence on me, and even let me develop in this research field.

Communicate with technicians in related fields. As a technical researcher, we often get close to the people in the industry who are engaged in this work, and I also try to understand what is happening in the industry. I often ask myself: Why do they do this? What can I do to change this situation? What important things have they overlooked? For academic researchers, these are usually opportunities to ask new questions and new projects. It should also be noted that many technical problems are not new strictly, and may have been studied long ago, so we can expand the reading scope and find new research ideas.

I have a document called "ideas.txt", which has been more than ten years. Whenever a new idea comes up, I will write it into this document. Maybe some ideas are not very good, but this habit is always good. Academic writing is also a very important basic ability. It is suggested to read 5- 10 before submitting to the journal. Learn more about the magazine and help communicate with other researchers in the community.

2. Process: Towards the goal.

You can find some of the best papers in the conference as your role models, and then try to make your research equally excellent. As a computer scientist, I have an algorithm:

Let bp = the best paper in the same conference?

In the past five years?

And my thesis & ltbp:?

Let why = Why bp is better than my paper?

My _ paper+= Why?

Submit my paper

Independent writing and multi-person cooperation

I usually let doctoral students write the first draft of their thesis independently, and before that, they will spend weeks or months fully discussing with me. Undoubtedly, academic writing is a very important basic skill. Being a writer is very important for academic career. The process may be painful, but in the end, the value of this pain will be proved.

When I find that there are still two or three months before the deadline, I usually make a completion plan for students' papers. For students in grade 12, it may be to give full feedback and guidance to individual chapters, and then students will rewrite them. At the same time, I will also make a lot of revisions and rewrites. For the students in grade five or six, I will listen to their ideas, and then ask them to write a "best paper they can write", and then I will revise it.

3. Within "-7" days? Deadline? rule

I used a "-7 days" for more than five years? Papers should be submitted 7 days before the deadline (but not the perfect version). If there is no formal deadline (such as rolling logs), then this rule applies to the day when we decide to submit in advance. This may be contrary to the way many scholars in this field work. They write crazily before the deadline, so I won't ask my students to do that. The paper doesn't have to be perfect: my completion standard is that I won't feel embarrassed to hand in the paper. It means that the research must be complete, the data analysis has been completed and the conclusion should be clear. Then we will see what we need to do in the remaining seven days.

The smell of wine is afraid of the depth of the alley.

If your research is great, you must learn to let more people know. When publishing the research results, I give students two suggestions:

You can send your work to the researchers who inspired you, and maybe there will be a reply.

You can write some media-oriented publicity articles about the research results, so that the media can further report your research.

In general, staying active on social media is really helpful.

1, internship experience

In addition, I strongly encourage my students to go out for internships for two reasons:

First of all, you can get the necessary economic income, because the salary that graduate school can provide is not high; Secondly, I think students should know what this industry is really like. Maybe at some point in your life, you find that you need to choose between academia and industry. Without some internship experience, it is difficult for you to make a decision. Of course, graduate school is different from becoming a real scholar, and internship is also different from being engaged in the industry for a long time, but in any case, internship is still meaningful for exploration.

Students in our field often practice in Microsoft Research Institute, Microsoft, Google, Facebook, Twitter, Airbnb, Reddit and other enterprises. In addition, emerging startups are also good opportunities, and you may undertake more diverse and rich job responsibilities. In the company, interns have a wide range of responsibilities, from pure research to pure development. If you have an internship place you want to go to, you can ask me. I may know someone there to recommend it.

Step 2 learn to rest

You don't have to work all the time to be a successful researcher. In fact, current research proves that working all day may be counterproductive. You may feel a "always available, always working" atmosphere in the academic circle, but I suggest you resist this culture. Be kind to yourself, no matter what. For example, I will have a good sleep, exercise, spend time with my family and find the boundaries between work and life.

The doctoral students in our group have six weeks of free vacation every year. At least I insist that you rest for at least four weeks. This period of time means no work emails, no work documents and no data analysis. Please note, however, that a trip to a conference cannot be counted as a vacation. If you want to take a long vacation, please inform me several times in advance. Of course, I'll try to be flexible.