Sunday, January 5, 2014

You Need Patience When Doig Research

If you are doing a PhD, you really have to observe and read a lot. This needs a lot of patience. You are less likely to succeed by starting to code once you decide on your topic.

Here is a good example of a research project that needs both dedication and patience:

http://news.cnet.com/8301-17938_105-57616458-1/dogs-do-doo-duty-in-alignment-with-earths-magnetic-field/

The article explains:

"The study shows a high level of dedication, with the researchers logging 1,893 dog-defecation observations using breeds ranging from beagles to mutts. The data was collected outside in open fields so dogs wouldn't be biased by routines established during regular walks. …"

Best wishes for a happy 2014, with both dedication and patience that you will need to graduate!

Sunday, September 1, 2013

What Does Your Adviser Say About You, While You Are Away?


This might be hard to know, but good to know if you can. If you are a postdoc, this is even more important because the chances of you landing on a faculty position depend heavily on the recommendation you get from the advisor.

A good boss will be critical of your writing style, and regularly point at things that you can improve, but will speak proudly about you with others. He will introduce you to fiends and fellow researchers, and help you to create a good network of connections. I have had two of these during my research career, both of them before I started my phd.

A bad one will butter you up once in a while, but  interesting things can happen when he talks about you with others. Here is one of my experiences.

I was a "research assistant professor" (an assistant professor hired for a short term using a research grant). My boss, the principal investigator of the research project, was a middle aged professor. Once we attended a social session at Experimental Biology 2011, to present a paper. I was the first author, and the boss wanted to attend because of the opportunity to meet researchers and experts.

On the day of the conference, we reached the conference venue together.  We reached the venue of the session, about 30 minutes earlier. The next step was to introduce myself to the session chair so that she does not have to worry, but she was not in and there was plenty of time for that. So, I went to the lounge to get some fuel (I mean, coffee :-p).

When I came back, the session chair was in and talking to my boss. So, I approached them and stood there waiting for my turn. My boss, who did not see me coming in, kept taking; "the paper will be presented by one of our post docs". "You mean him?" asked the session chair, pointing to me (i guess like all good session chairs, she did her homework and googled the home pages and photos of the authors).

I put on a poker face, and the boss turned red. "Actually, he is an assistant professor." Then the session chair started taking to me, before we all sat down and started the session.

In fact, the prof used to be a generally nice person. However, he created a startup attached to this project, and seemed to think that researchers and students in the project should help his startup rather than focussing on research. I guess this made him a bit grumpy. I could also be that he wanted to use the contacts for the benefit of the startup, and avoid others from contacting me.

I wasn't happy to know that I am regarded as a postdoc, when i was working hard as an assistant professor. I am sure it was no mistake, given the short time it took to correct the mistake and his facial expression. For me, I am not the type who would take this kind of thing face to face. Instead, I decided to change my job soon.

Wednesday, August 21, 2013

Work-life Balance: Professors' Comments

Some professors were not happy with the content of my previous post, about "work-life balance".

"It should be Work-Work Balance! Grad students are not supposed to have a life.", argued one professor.

Another professor was more agreeable. "Of course grad students should maintain work-life balance. Work is what counts towards their graduation, and TA pay. Life is what they do to help us (their advisors), from reviewing papers to troubleshooting our computers."

Seriously, professors are skeptical about work-life balance advise to grad students. "They already have a lot of free time, compared to industry workers. This sort of advise only makes them spend even less time on research!", grumbled another.

Talking about my own experience, my boss at the time was not amused when I told him that I attended a work-life balance seminar (he was attending the same conference, and wondered where I was during the lunch time). He thought that kind of seminar is not relevant to the conference.

Industry people are much more tolerable with this kind of stuff. The research scientist who supervised me during my first internship (in Japan) said "I do not work on weekends, and do not expect you to do so. You can go out and have some fun."

The point is; you are most likely not making your advisor happy, by telling him that you are focussing on work-life balance. But you don't have to tell him, just do it!


Sunday, August 11, 2013

Doing Favors


Research community is seen by outsiders as "fare" and unbiased. After all, their work has been assessed by peer review for the past few centuries. However, in practice, there is as much favoritism in the research community as anywhere else. If your prof is well known, you will be enjoying this situation; sometimes fully unaware of it. Otherwise, you might end up frustrated.

I was a publication co- chair of ACM Multimedia 2012. My main job was to handle the proceedings, but I was involved in a few other tasks including conference registration. Here is one of my experiences with paper registration.

As most of you are aware, authors of accepted papers have to pay full registration fees per each paper, and make sure that at least one of the authors attends the conference to present the paper. Some authors try to save money by not attending the conference, because once the paper is in the proceedings, it is permanently in their publication record. This is called a "no-show", and some conferences threaten to blacklist authors who are involved in no-shows.

Immediately after the list of accepted papers for ACM Multimedia 2012 was published, a student from Malaysia contacted us. She had a paper accepted as a poster, but her university was unable to provide enough funding for both registration and travel. The conference was accepting travel grant applications, so we asked her to apply.

Two weeks passed, and we were approaching the registration deadline. The results of the travel grant applications were still not announced, and the student got worried. She asked whether she will be blacklisted for no-show if she registered the paper anticipating travel grant support, but later could not secure enough funding.

Since none of the big guys - old profs who were the decision-making chairs- responded, one of the younger chairs raised the matter in our mailing list. One of the top 10 responded with the textbook reply: if you don't attend, it is a no-show.

Okay, so far so good. But a few hours later, the moment of truth came. A professor who had four papers (let's call him Prof. N.) wrote to one of the top three (say prof. Z) asking for a discount for his registration. And Prof. I wrote to the mailing list, supporting this request.

We strongly objected, because there were more than 25 authors with 3 or more papers, and more than 50 with 2 or more. To make matters worse, all the other authors with 4 papers had already paid. "We will go bankrupt if we start giving concessions", argued one of the web chairs. Prof. Z retorted "a conference is only as good as its attendees, and attendance of this person is important for the conference". But I knew the real reason for the request; prof. N was slated to be one of the big guys in ACM Multimedia 2013 organizing committee, and was a good friend of Prof. Z. The argument was continuing over several emails.

We, the younger chairs, thought this a good time to raise the issue of the student from Malaysia. One of us sent an email to the mailing list, reminding that we haven't advised her, and asked which options should we offer to her. "This, in my opinion, is more important than giving a discount to an author with multiple papers, as discussed in another thread", I shot back. This angered Prof. Z, and he tried to defend himself by saying that the student is a one-time author (implying a noob) who is trying to not pay, unlike the valued prof. N. But a few others, including a program chair from the industry (thanks, R) supported the student. In the end, one of the more sensible out of the older chairs advised us to let the stiudent register first and look for alternatives later. "We have to accept the realities of financial problems, when it comes to students with less funding", he summed up.

Throughout this time, I was unofficially in touch with the student, asking her not to get discouraged. She was thankful to the younger chairs who were fighting on behalf of somebody they never met. We won this time, but it left a bad taste in everyone's mouth (for completeness, let me mention what happened to Prof. N's request; Prof. Z approved it as "an exception", but I am not exactly sure how much was discounted).

Some of the old guard might say that favoritism is everywhere in the society, and the research community is no exception. I think that is a pretty lame excuse for those who are entrusted to spend large amounts of public funding with relatively loose conditions, and evaluate their own work.

So, the message is: if you are a grad student who has a known advisor, make sure that you play fair. If you are not, make sure you play fair when you graduate, if you happen to stay in the research community.

Sunday, July 28, 2013

Don't Steal Titles!

When you decide on a title for your paper, it is important that you check whether the title is already taken. Also check if your paper title contains the title of another paper; there is a high chance of this happening nowadays, since most papers have long titles.

With Google, or any other good search engine, this should take only a minute. Omitting this important step can cause the following problems later on:

1. When people search for your paper, they will end up finding a different paper.
2. The reviewers might think that you are using the title to steal someone else's reputation; this is bad for you.

Just in case; if you ever thought of using a title similar to that of a good paper, just to attract readers, don't do it! That is totally unethical!! And more, people will soon find out.

Recently, I searched for the paper:

Marziliano, P., Dufaux, F., Winkler, S. & Ebrahimi, T. (2002), "A no-reference perceptual blur metric", In Proc. International Conference on Image Processing, Vol. 3, pp.57-60, 2002.

And found:

Luhong Liang; Jianhua Chen; Ma, Siwei; Debin Zhao; Wen Gao, "A no-reference perceptual blur metric using histogram of gradient profile sharpness," Image Processing (ICIP), 2009 16th IEEE International Conference on , vol., no., pp.4369,4372, 7-10 Nov. 2009

One can argue that the authors of the second paper were unaware of the first. But the list of references in the second paper includes:

P. Marziliano, F. Dufaux, S. Winkler, et al, "Perceptual blur and ringing metrics: application to JPEG2000", Signal Proc.: Image Commu., 19(2), pp. 163–172, 2004.

This is a later work by P. Marziliano, and less related to the topic compared to the original paper. So, no one can blame anybody who suspects that the title was "copied".

If you do something like this by mistake, it will embarrass not only you, but also your co-authors. Especially your advisor who won't spend much time checking this kind of detail :-p

Saturday, March 2, 2013

The Authors and Co-authors

If you are a grad student, I am sure that you are reading a lot of papers (okay, I mean you at least starting to read many, and then give up on the unreadable) :-p. You might have noted that the recent papers have more authors than the older papers. At least this is the case for Computer Science research.

Let's take a more objective look at the number of authors per papers. I selected ACM Multimedia, one of the most famous multimedia conferences. The 20th Annual conference was held in last October, so there is a good time span to observe trends.





You can see that the number of authors per paper has increased quite fast, in the past 5 to ten years. The trend is the same for other multimedia conferences such as ICME. Here are the main reasons for the increase.

1. Just like you need publications to graduate, those above you need publications for different purposes. In some universities, professors get a pay cut if they fail to publish a given number of publications. Non-tenured faculty need papers to get tenure. Postdocs need more papers so that they have a better chance of getting in to the faculty. Funding in academic departments are allocated by looking at publication records. I guess you now know why everyone above you want you to publish.

2. Things at research institutes are worse. Most researchers are on contracts nowadays, and they have to publish to keep their jobs. Some institutes (example: I2R, Singapore) require researchers to publish only in conferences recommended by them. You might feel very happy that you got an internship at that famous research lab, but a greedy-eyed researcher might be waiting there, wanting you to publish from day one.

Okay, people needed publications not just 10 years ago, but before that as well. So what has changed? Funding for research has reduced quite rapidly during the last decade. If the funding increased by the numbers, then the number of people who used the funding has increased. So, everybody wants more papers with less money. The solution: put more names in one paper!

One young researcher with whom I work for a short while told me "if we put each other' name in our papers, we both will have publication records that are twice as long." Sounds mathematically correct, but ethically, wrong. When I pointed that out, he responded "I did not create the system, I am just trying to survive in it."

3. In the early days, researchers got funding from "funding agencies", and mentioned the agencies in the "AcknowldegmentS" section of the paper. But since recently, some research institutes are using money to "buy" papers. When you get funding from such research labs, they assign you one or two collaborators, who might (or might not even) have an occasional Skype meeting with you. You put the names of these collaborators on the papers. Arguably, it is a good thing to have a collaborator from a large research lab. But it is not correct to put names on other people's papers just because of money.

What these people don't realize is that the industry is watching. Very soon, there will be many researchers with huge publication records (okay, there already are). Once credibility on the publication system is lost, it will be very hard to restore.

So, whose names should be on papers? Ideally, the names of thoe who did the work. The supervisor's name is fine, because he/she guides a student who is learning to do research. Proofreading, being a friend, providing money, etc., do not count as valid reasons.

But then, you as a grad can't do much about the names your supervisor inserts in your paper. At least try to put your name as the first author, if you did the work. In many universities, the professors are not allowed to be first authors when students write papers. And when you graduate, be more ethical if you still publish.

Sunday, November 4, 2012

Moving to the Industry

I moved to an industry position starting from this month. This will allow me to post more freely regarding some practices of the academics (both good and bad).

I also served as the publication co-chair of a major multimedia conference. With what I have witnessed, I have more things to tell you :-p