Some Advice for Graduate Research Assistants

Minh N. Do

January 26, 2009

In a group discussion about the experiences of graduate students in signal processing at the University of
Illinois at Urbana-Champaign (UIUC), Prof. Pierre Moulin listed ten qualities of an ideal Ph.D. candidate:
intelligent, creative, proactive, driven (motivated), confident, good communicator (oral & written), indepen-
dent, hardworking, well organized, and able to work with others. People in that meeting agreed that while
some of these qualities we are born with, fortunately many other qualities can be obtained through practice.
In this memo, I offer some advice on the three most important qualities, using my personal experience
when I was a graduate student. These are: writing skill, presentation skill and being proactive. I then suggest
a concrete action plan to regularly practice these skills. Any comments or suggestions are always welcome.

1 Writing Skill

Writing skill is an essential quality of a graduate student. While there are many other factors, the productivity
of research nowadays is measured mainly in terms of the number of papers in high quality journals1 . These
products will ultimately be the result of an excellent writing skill.
So how can one improve his/her writing skill? Following are several activities that I found useful:

• First, you need to have some good books on writing. These books typically contain the basic rules in
writing, as well as plenty of good tips on how to write well. Here are some of the books that I use
frequently and keep within a close reach whenever I write.

1. Higham, Handbook of Writing for the Mathematical Sciences, 2nd ed., SIAM, 1998.
This handbook is highly recommended and should be on the reference shelf of any research
student. The book covers a handy set of rules and provides advice on technical writing. It also
overviews the entire publication process, which is invaluable for anyone planing to publish in a
technical journal.
2. Strunk and White, The Elements of Style, 4th ed., 2000.
This little book is a classic and “should be the daily companion of anyone who writes.” It
presents a concise set of rules and principles, together with examples, about English writing.
Elements of Style is available on-line at http://www.bartleby.com/141.2
3. Zinsser, On Writing Well, Collins, 2006.
Another classic guide for nonfiction writing.
1
In our field, the most most respected journals are the IEEE Transactions on Signal Processing and IEEE Transactions on Image
Processing.
2
From www.bartleby.com you can also access to other essential writing resources, with their search facility, such as Amer-
ican Heritage Dictionary of the English Language, and American Heritage Book of English Usage.

1
 4. Alley, The Craft of Scientific Writing, 3rd ed., Springer, 1996.
This book offers excellent discussion and useful suggestions about how to bring your idea into
strong writing.
5. Dupre, BUGS in Writing: A Guide to Debugging Your Prose, 2nd ed., Addison-Wesley Pub Co,
1998.
Another good book that has writing advice and examples.

In addition, a good English dictionary is indispensable. Since I do most of my writing directly on

computer, I find computer-based dictionaries are very useful, because of their search facility. Com-
puters on the UIUC network have access to the Oxford English Dictionary via:
http://dictionary.oed.com/entrance.dtl. There are also many other dictionaries avail-
able on-line, such as Merriam-Webster Dictionary and Thesaurus at http://www.m-w.com.

• As you would learn to program best by studying and imitating programming examples, good technical
writing skill requires reading lots of technical papers. If you plan to write papers for a journal, say
the IEEE Transactions on Image Processing, then you need to read many papers from that journal
to get an idea about the common style, level of details, structures, length, format, etc of its papers.
Start a habit to note down phrases, constructions in papers/books that you find useful so that you can
apply them later to your papers. Doing so you could avoid being plagiarism. Never copy a whole
sentence from other author(s) into your writing unless you want to quote. In addition, try out to see
what writing constructions work best for you. In the end, you have to develop your own writing style.
Note that even in journals, not all papers are well-written. When reading either a strong or weak
writing, try to figure out why so. Remember these observations when you are in the other role as a
writer. A golden rule in writing is always keep the audience in mind.

• Practice, practice, and practice! The best way to improve writing skill is to practice it regularly – a
more specific action plan about this will be followed shortly. A well-known writer even said that you
should treat writing as “taking a dog out for a walk !” That means you need writing something (even
as little as one paragraph) every day to keep up with the practice.

• Revise, revise, and revise! Revision mean “re-seeing.” Remember the writing task does not end after
finishing the first draft. Revision is the key for good writing. Revision is not just about spotting
grammatical and mechanical problems, but it is also about “seeing the big picture and the overall
effect.” Chuck Guilford offers the following excellent advice about revision on the Paradigm Online
Writing Assistant Web site (www.powa.org).

Knowing you’ll revise, you can relax and speak your mind early in your writing process.
Since your words can be changed later, you won’t worry about writing the perfect first
draft.
That’s one important difference between writing and speaking. When you speak, you get
only one chance. Whatever you say, appropriate or not, has been said. Maybe that’s why
we have the sayings, “Think before you speak,” and “Make sure brain is in gear before
engaging tongue.” If you write something inappropriate, however, you can change it, or
even tear it up and throw it away. With writing you get a second chance, or a third or a
fourth. And each time you revise, you find new potential in the evolving text.
Too often, inexperienced writers don’t see this potential. They are too careful and self-
critical at the start of a project and too easily satisfied toward the end. Having agonized
through a first draft, they quickly check for grammar and mechanics and consider them-
selves done.

2
 More experienced writers usually do just the reverse. Early on, they work at discover-
ing what to say, getting their ideas out onto disk or paper. They write quickly, accepting
chance discoveries, trusting hunches and gut-feelings, willingly making mistakes. Gradu-
ally, though, they feel a need to look back over their work, to ask whether it makes sense,
how their readers will respond. Thus begins the process of revision.

• Try to get help and feedback on your writing, especially if you are non-native English speaker. In that
case, it is ideal if you can find a native English speaker who would be willing to proofread your writing.
For this, the Writers’ Workshop at UIUC (http://www.english.uiuc.edu/cws/wworkshop)
offers an excellent resource. Also from its Web page, you can find many links to other resources on-
line.
Print out your writing in double-spaced format and ask someone to comment on it using a color pen.
Study and keep these notes. You will soon discover the common problems that you have in writing,
such as incorrect use of prepositions or omission of the ending ‘s’. Work hard on fixing up these
problems by yourself in revising. Also, remember to thoroughly revise a writing by yourself at least
once before giving it to someone for comments.

2 Presentation Skill

Presentation skill is required in all forms of oral communication, ranging from discussing research prob-
lems with other researchers to presenting papers at conferences. Apart from regular meetings with your
thesis advisor, you should frequently discuss your research with your peers, other faculty, and even people
outside your area to get feedback from different perspectives. Good presentation skill will maximize the
benefit from these discussions. During your time as a graduate student, you will have plenty of chances to
present research to your colleagues, faculty members (some will be on your thesis committee), audience at
conferences (some could be your future employers). You should seize these occasions as great opportuni-
ties to demonstrate your research and communication capabilities. Most of the leading researchers are also
well-known for being excellent presenters.
Following is some advice for improving presentation skill.

• If you are a non-native English speaker, you should consider taking some oral communication courses.
Even native speakers benefit greatly from public speaking course (business people often do this).
The University of Illinois offers several ESL (English as a Second Language) courses. Check the
International Student Office for more details about these courses.

• Prepare for a Teaching Assistant (TA) job. I consider taking TA appointments is an important part of
the PhD education. This is of course essential if you plan to get an academic job. But also in many
other occupations, the ability to explain difficult concepts to other people is essential. If you are a
non-native English speaker, you need to pass the SPEAK test to be qualified for a TA job at UIUC.
Find out the requirements and consider taking the SPEAK test as soon as you can.

• Like reading is an excellent way to improve writing skill, attending seminars is an excellent way
to improve presentation skill. There are plenty of seminars on the campus during the semester and
you should benefit from these seminars. Apart from our weekly group seminar, the most relevant
ones to us are the DSP seminar and the ECE seminar at UIUC. Come to these seminars to learn new
developments in closely related fields. Also observe the presentation techniques of speakers, and find
out what work and what don’t. You will find the common mistake in many talks is that they are too

3
 complicated, too detailed, and thus loose the audience and don’t show the big picture. Remember to
avoid this mistake when preparing for your own talks.

• Seek opportunities to practice your presentation skill. Talk often to your peers about your research.
If you can present your research/idea in interesting and accessible ways then people would keen to
hear about it and more likely to give useful feedback. Volunteer to give talks in seminars. The most
effective way is to organize weekly seminars with a group of graduate students who shares common
research interests and each person in turn would select and present a seminal paper in the field.

3 Being Proactive

Being proactive means “acting in advance to deal with expected difficulty”. Working on a PhD thesis is
certainly a very difficult process. It requires formulating the right problem, finding feasible solutions, and
conducting simulations to confirm the theory/intuition. While the thesis advisor can provide initial direction,
the PhD candidate has to be in charge and lead the process.
New graduate students often find it is difficult to be proactive in research, especially in coming up with
new research ideas. One way to get a few good research ideas is to generate lots of ideas.

• Read widely. Reading, in my opinion, is the best way to learn. However, to get the most benefit out
of reading, you have to read with a critical mind.
First, for reading there is a vast literature out there on any topic, so you need to be selective of what
to read. Prof. Doug Jones offers excellent tips on using the library for research at
http://www.ifp.uiuc.edu/˜jones/library.html. He suggests to start general, then
specific. If you are new to a field and that field is mature enough to be covered in books, then you
should start from there. Library search facilities will help you to find these books. For more spe-
cific or current topics, you may only find the desired material from papers in journals or conference
proceedings. Fortunately, in our field most of these articles would be published by IEEE, and thus
the IEEE Xplore website (ieeexplore.ieee.org) with its search facility is an indispensable re-
source. Another resource is search engines like Google Scholar (scholar.google.com) or Scirus
(www.scirus.com), since many people make their work available on-line. For example, if you are
investigating “directional filter banks (DFB)”, then these search facilities could provide answers to
some crucial questions like “what has been done with DFB?”, “what are the latest developments in
DFB research?”, or “what are applications of DFB?”.
Next, for critical reading the most important thing is to read with an active mind rather than with an
passive mind (i.e. only absorbing the information). Keeping make questions like why did the author(s)
do this? How did they come up with this idea? What is the key intuition behind this technique? How
could we do it differently? What happens if we change the assumption to...? This is a great idea, may
be I can apply it to my problem (could be in a completely different field)?, etc.. Write all of these
comments/ideas down. Don’t just trash an idea right away because you “feel” it is not feasible or
invalid (e.g. based on assumption that you don’t have enough background in that area). First, list all
of your ideas; then revise and modify them regularly later.
Another good tip is to read and re-read some of the classic papers in your research area from time to
time. These papers are often very stimulating and inspire new research ideas.

• Get insight from data and experiments. If you find an interesting algorithm or method, go ahead and
implement it and test it on your data. For example, when I learned about ridgelets and curvelets,

4
 I wrote nearly a hundred Matlab programs that varied different aspects of these transforms. I then
printed out the results, stuck them on the wall, stared at them every day, and then eventually ideas for
improvements and novel methods came up...

• Talk and seek comments/feedback from other people. Doug Jones, on his library search Web page,
suggests the following excellent tip:

Tip #5: Ask the experts: At the University of Illinois, you are blessed with one of the
largest and most knowledgeable body of scholars on Earth, including not only the faculty
but also staff, graduate students, and undergraduate students. Ask around, find out who
the experts are on the topic of interest, and go talk to them. Most scholars greatly enjoy
sharing their knowledge with others who have a genuine interest, so take advantage of this!
However, keep in mind that you need a basic understanding of the area to frame the right
questions and to understand the answers you get, so experts are not a complete substitute
for written resources. Use them like the more specialized material mentioned in Tip #4;
prepare yourself first to take advantage of their expertise. But, don’t be too shy about
approaching them, either!

4 Others

There are many excellent resources for graduate students and researchers on the Web. Here are some that I
found very useful:

• How to Succeed in Graduate School?

http://www.cs.indiana.edu/how.2b/how.2b.html

• Excellent advice about graduate school life.

http://www.cs.unc.edu/˜azuma/hitch4.html

• Research Resources. “Useful tips on how to succeed in graduate school and your subsequent research
career (somewhat slanted toward signal processing).”
http://www-dsp.rice.edu/˜richb/resources.html

• Graduate Student Resources Page. It contains a wealth of information.

http://www-personal.umich.edu/˜danhorn/graduate.html

• Collected Advice on Research and Writing.

http://www-2.cs.cmu.edu/afs/cs.cmu.edu/user/mleone/web/how-to.html

• PhD - First thoughts to finished writing. A detailed discussion on how to get a Ph.D!
http://www.uq.edu.au/student-services/linkto/phdwriting/

• Networking on the Network. “A guide to professional skills for PhD students.”

http://www.acm.org/crossroads/xrds4-4/network.html

5 Action Plan: Weekly Report

In order to practice these above mentioned skills, I recommend the following activity for graduate research
assistants (GRA) in our group. Every week, each GRA should write a one-page report summarizing the main

5
research findings of the week and providing the working plan for following weeks. We can think about these
weekly reports as journals of research activities. Exemption for this rule is when you are writing something
else like a paper or thesis and you can turn in a draft for comments instead. The benefit of writing weekly
reports is twofold. First, it serves as a bookkeeping of your activities and new ideas (e.g. after reading a key
paper), and provides a write-up that can be used later in papers, your thesis, etc. Second, it offers a regular
writing exercise. Thus, the reports should be written in the style of a scientific paper.
Following is an example format for the weekly reports:

Statement of Problem/Abstract: A concise description of the current research problem/paper(s) that you
are working on.

Main Findings: Describe your main findings of the week. They can be a summary in your own words of
the relevant results from research papers that you read during the week; new ideas/suggestions for
improvement; a discussion of a new set of experiments that you conducted and the results that you
obtained; new theorems or conjectures.

Future work: What would be the next steps based on the current results/findings. If possible, sketch a time
table for these steps.

Appendix: If necessary, attach copies of plots, proofs, etc.

6
A Appendix: Using LaTeX for Writing Papers

If you have not done so, you should learn how to write technical documents in LaTeX as soon as possible.
Apart from producing very professional looking documents, especially with mathematical formulas, LaTeX
makes it very easy to format your papers to meet guidelines of conference proceedings, journals, thesis, etc.
Also, you can easily change from one format to another, for example: conference paper → journal paper →
thesis. Moreover, since most of the researchers in our field write in LaTeX (as I do!), if you also do the same
then it becomes very convenient to combine writing with other people in co-authored papers. Finally, since
you would write using a plain text editor, LaTeX helps you to focus at the content first instead of worrying
about the format.
Although having a LaTeX book on your desk is crucial (the recommended one is Guide to LaTeX2 by
Kopka & Daly), for start a short introduction is sufficient to get you going. For this, the on-line document
The Not So Short Introduction to LaTeX
(http://ctan.tug.org/tex-archive/info/lshort/english/lshort.pdf) is excellent;
it can get you start writing a LaTeX paper in 87 minutes! Also, ask friends to give you some of their LaTeX
files to use as a template. Paper templates for conferences and journals are often provided on their websites.
The LaTeX system should be already installed on your campus computers. For an Unix machine, this
is always the case. For a PC, you might need to download and install a free version of LaTeX for PC
which is called MikTeX (www.miktex.org). For writing LaTeX files, xemacs in Unix and WinEdt for
Windows are the most popular options. Again, these programs should be already installed on your campus
computers.
You will very soon have to include graphics into your papers. The on-line document Using EPS Graphics
in LaTeX2e Documents contains most of things you need to know about this. Both this and the above men-
tioned on-line documents are available from http://computing.ee.ethz.ch/text/latex.en.html.
For drawing figures, xfig is a simple but very effective program. I keep all of my figures, drawn by
xfig and then exported to EPS files or generated from Matlab programs, in a directory that can be shared
by several papers and presentations.
Finally, for citing references, BibTeX is a powerful tool. It helps you to maintain a database of pa-
pers/books when you read and survey the literature, then use in LaTeX documents, and also share with other
LaTeX users. Here is an useful convention in naming the bib-item: it starts with the name of the first author,
followed by the last initial of other authors (if any), then ’:’, and the last two digits of the published year.
For example:

@BOOK{BurrusGG:98,
author = {C. S. Burrus and R. A. Gopinath and H. Guo},
title = {Introduction to Wavelets and Wavelet Transforms: A Primer},
publisher = {Prentice-Hall},
year = {1998}, }

Following this naming convention will help you to cite a reference in your paper without checking the
BibTeX files.