Tips to Graduate Students

Graduate School 101: Tips to My Graduate Students

Over the last number of years, I have had the honor to work with a group of excellent graduate students. The everyday interactions with the students have sharpened my understanding on what it takes to be successful in graduate study and how to make most out of it. I summarize what I see as important pieces that put together a successful graduate school endeavor.

ABC's of Research

One of the greatest experiences a graduate study can offer is the opportunity to conduct research. By definition, research is a process of creating and developing new knowledge or solutions to unsolved or new problems. This very nature makes the process of research very dynamic. Across many science and engineering disciplines, there are few common key characteristics that make one successful in research: the abilities to learn, understand, communicate, develop and innovate. More broadly, these are also the central qualities that make one competitive in today's global economical and technological competition.

How to start a research endeavor? Research activities start from a solid understanding of the current state-of-the-art. Simply put, if you don't understand, you cannot innovate. This early stage of research usually consists of reading textbooks and literatures. But don't just completely follow what's already out there. As you read, also try to be a critique by asking: why is this? What’s wrong with a simpler approach? can we do better? Do not rush to settle a nontrivial question by comforting yourself with a superficial answer. Enjoy this questioning stage because it lays out a hotbed for fresh new ideals. If you don't criticize, research opportunities have already slipped away. Put in a different way, if you become a slave of existing literatures, you've lost the mentality for being a researcher.

Problem-solving skills are for sure highly emphasized here. In reality, problem solving may take place in a richer manner than what you may have thought about. In research, you don't necessarily have a fixed problem to solve. As you develop more understanding of a subject matter, you would have a better picture about what the real issues are and where you can make contributions. So, defining, redefining and refining your research targets are integral parts of the problem solving process.

Once you have a good understanding of your target problem, don't stay in the planning stage forever. Start to put your hands on experiments. You may start from a simple idea and see how it works. Do some basic proof of concepts. As you start to see more reality through experiments, believe me, you will run into unexpected problems. So, what will happen is that you will be bothered with more headaches. But this is not bad. At the same time, more solid understandings and new perspectives will come to you and in fact you will be actually in a better position to achieve your goal. The complete research process consists multiple of such thinking, experimenting, re-thinking iterations. How well you embrace this dynamic process will determine how much you will accomplish in the end.

If you are a beginning graduate student, you are doing a great job if you can follow nicely the basic principles and put the ideas of your advisors and collaborators into a working solution. If you are not so fresh anymore, you should start to fully engage in the process of problem definition and solution proposal. If you are not doing these more creative activities yet, you may have already fallen a bit behind. Anyway, the best way to become a good researcher is to start researching. Act and get into this process of new knowledge creation now. Even you may start from a blurring concept at the very beginning, you can be very much amazed by what you will accomplish in the end. When this happens, all your efforts pay off.

How to write a technical paper

Technical publications are primary means to disseminate your new findings to the research community at large. You have learned great names from great papers. Likewise, your readers will also develop an image of you through your work. People often say "only when you start to write, you realize how sloppy your thinking was". So, the process of writing is also a process of refining one's thoughts. Surprises and new perspectives often surface. So, enjoy writing and keep practicing! Some basic tips for pulling off a paper are:

Abstract:
In concise language, you basically convey the following: what's the problem you try to solve? Why is the problem relevant? What's the proposed approach? What have you achieved by your approach? Use a couple of sentences to address each of the above items.

Introduction:
The introduction can be thought as an extended version of the abstract. First, you define the research problem that you want to address; you may motive the need for the intended work by providing application context, technology trends etc. Next, you may provide a technical context by reviewing related literatures; a thorough literature review is very important since it demonstrates your understanding of all relevant issues and provides a coordinate system for other people to judge your work; this is also a good way to acknowledge other researchers whose contributions have motivated your work.  Then, you shall summarize the basic line of thinking, working principles and main steps of your proposed solution; you shall contrast your approach with existing work by pointing out its novelty. Finally, you describe what results have been achieved and highlight your experimental findings. Essentially, here you want to clearly convey to people: what problem you try to address, what is new in your approach or where you've made contributions, and finally what results you have achieved.

Background:
When appropriate, you may use a section to provide required background knowledge for the proposed work. For instance, your work may be motivated by some other people's work, which is not universally known. Then, you may provide an overview of this prior work to provide a basis for the readers to understand your new contributions.

Overview of the approach:

When there is a need, you can make your ideas more accessible to other people by offering a roadmap-like overview. By doing so, you can bring a clearer big picture to the readers early on and reduce the "risk" of having them lost in the following sections. You should properly balance between this section and the introduction.

Main body:

Next, you describe the key steps of your proposed method in a very detailed fashion. You should do this in a very clean and logic way. Some of tips are the following. a) Well organize your paper by nicely putting things into sections and subsections etc. b) Pick up informative section/subsection titles so the readers can clearly see the flow of the materials. c) Transit smoothly between sections and avoid sudden jumps. d) Each section/subsection should have a clear purpose to serve. For instance, you may talk about a specific issue in a section. You can divide this section into multiple subsections and use each subsection to cover a key aspect of the issue in question. You should approach any of these aspects in a clean and logic way: first, state what you are trying to accomplish; then, say a few words about the difficulties in achieving your goal; finally, describe how you address these difficulties. When appropriate, you shall also state clearly all the assumptions that have been made in your approach. e) Try to use clean and polished languages. Avoid using excessively long sentences -- break them into more manageable pieces. Also avoid extremely long and flat paragraphs.

Experimental results:

Experimental results are essential parts of a paper. They should never be treated lightly. Before presenting any specific experimental findings, you should clearly describe how your experiments are set up. This will provide a basis for readers to interpret your findings and judge whether or not your experiments are set up in a meaningful way in the first place.  Detailed experimental results shall be provided to verify all your proposed ideas.  You should not present any "new" idea that is not backed up by experiments. Think about effective and interesting ways to demonstrate the power of your new approach. Provide thorough experimental results by playing with multiple test cases with varying complexities and characteristics. Examine key tradeoffs experimentally. Be aware that the way in which the experimental results are presented is also a reflection of your understanding of the problem. If you really know what truly distinguishes your work from others, you shall know how to reflect it in experiments. You should not only present your experimental data, but also interpret them. Figures and tables shall be used effectively. Pay attention to captions of figures and tables. Make them informative and specific. For example, a figure caption like "Validation of the proposed method on circuit 1" is non-specific and doesn't carry much information.

Conclusions:
Here, you basically want to have a recap of what have been done in your work. You may also point out your future work.

How to give a technical presentation

Presentation skill is an important communication skill and plays a big role in things like conference talks, thesis defense and job interviews. Be aware that doing a good talk is much more than just speaking good English. In some sense, how well you can present is a total reflection of your professional quality. Within minutes of your talk, your audience may develop a clear picture about you: how thorough is your understanding of the problem? Can you nicely convey the key messages of the subject matter? How energetic and confident are you? What’s your style? Can you present yourself to the audience in a polished and professional way? Can you anticipate the audience's concerns and handle questions intelligently? No wonder a 45-mintues or one hour presentation is always a central piece of any serious job interview.

How to give a good talk? First, you should understand there are good reasons for why we should do a good job here; this is not just an extra homework assignment you have to do. Second, before you can get your audience excited about your work, you should feel excited about it yourself. You should have the passion to come out to talk about it and share your experiences with others. Next, you need to be well prepared. The preparation of your presentation slides can follow some of the tips for writing a paper. If you are doing a conference talk, your may use the same paper organization to structure your talk. You shall be benefited from the interactive features of the presentation to tell a story in an interesting and accessible way. Some of the tips are:

a) Before you prepare your presentation materials, first think about how to present your ideas in an intuitive and interesting way. If you are giving a timed talk to a general audience at a conference (instead of a closed-door internal discussion), you should not expect throwing lots of convoluted details at your audience would have a good chance to succeed. It's important to provide intuitions. If you don't make your presentation interesting and accessible within the time limit, most of people will get lost and lose their interest very quickly.

b) When you start, first spend a couple of slides to define the problem context and motivate the audience.

c) Always have an outline of the talk so people will know what you will talk about throughout the presentation.

d) For most of cases, you should review prior work. Add references on the slides to provide pointers to these works.

e) When you go into the specifics of your approach, be clean and logic. For each issue, talk about "what you want to achieve", "what problems you have to address to achieve your objectives" and then "how you address these problems".

f) Use a professionally looking template and layout scheme. Be careful about your slide style.

g) Don't put lots of stuffs (pictures, words) onto a single slide to make it super busy. It's difficult for people to track a busy slide.

h) Don't fill a slide full with texts. People won't have the time and energy to read a full page of text. The bullets on each slide should be very short sentences or just keywords. Remove redundant wording as much as you can to make your slides clean. For example, "As you can see, we have achieved great results" may be shortened to "Great results achieved".

i) Font size shouldn't be less than 18pt. Avoid any overuse of colors. Having many different colors on a slide will make it very noisy. If you choose to use a particular color to highlight certain keywords, stick to the same color throughout the presentation.

j) Properly use the animation to make your presentation more interactive and flow better. If you have many equations on a slide, use animations to show them one at a time. Throwing many equations to the audience all at once can easily become too overwhelming.

k) Practicing the talk ahead of the time can be very helpful. Only after some practice, you can have some an idea of the issues in the slides. Then, you know what to do (add or remove materials) to make your slides work better for you.

Why should I show up in a seminar and ask "stupid" questions?

Attending seminars is an effective way to look into other people's work and broaden one's horizon. Within a time span of a few ten minutes or one hour or so, speakers may help you develop a basic understanding of a new territory and walk you into a new field. Showing up in a seminar is indeed a very efficient way of learning new things. It may save days, weeks of literature review effort otherwise would be required. Take the advantages of having an expert (presenter) within your reach in a seminar. A face-to-face interaction with an expert can provide you valuable information which are hard or even impossible to get from literatures. Ask questions. There is no such thing as a "stupid" question. Remember, when you ask questions (even "stupid" questions), you are grabbing more from the time spent on the seminar. So, this is indeed in your "best" interest. Asking questions is also a way to acknowledge the speaker since it shows your interest in the presented work. Anyway, if you keep asking questions, sooner or later you may notice your questions actually become "smarter". This is another benefit of doing so -- it sharpens your thinking and improves your interactions with others.

Don't expect to completely follow all seminars. It's perfectly fine if you get completely lost sometimes. Don't expect an immediate return from attending seminars. If you keep participating in various seminars, over a longer term, you will develop a fuller picture of the technical world at large. At some point, you may actually see ideas coming from diverse fields do connect! Additionally, through the successful and not so successful talks given by other people, you can learn how to do a better talk.

So enjoy a one-hour discursion from your everyday work once in a while. Welcome a change of perspectives a seminar may bring to you. It's worth your investment.