CS PhD Admissions & Publications

[President]

Hi guys!

Long time lurker, first time poster:D I usually try to read through all of the relevant posts that I could find about a topic prior to actually asking a question, so here goes! I read how a lot of the admissions pages on CS PhD websites try to encourage applicants to apply directly to the MS/PhD program, rather than to pursue a M.S. beforehand.

1. My question is, would pursuing a M.S. ever be a disadvantage to gaining admissions to a top program?

2. If it depends on the school, could someone please point out which school has this type of attitude towards those who have already obtained an MS?

I'm worried that adcoms would expect more from those who have already gotten an M.S.( that is, publications on a beautiful journal or conference). This brings me to my next set of questions:

1. Are M.S. Computer Science students pursuing research usually successful at having their research published on decently ranked conferences/journals? Or is publishing actually more difficult than it appears to be?

2. Does it really take a year or more for a journal to be published? (I know the preferred venue is conferences, but I really like writing longer articles, and they supposedly have higher accept rates).

3. Is it common to have a semester project in a graduate CS course to be published? I ask, because some CS courses involve conducting actual research... I'm wondering if I can have a Master's thesis on a journal + a few conference papers derived from graduate level CS coursework.

I honestly have more questions after reading through this forum, but I really don't want to write a book! Your thoughts would be thankfully received, and I, from the bottom of my heart, will return the favor to this community once I educate myself about this stuff:D

Edited April 29, 2011 by President

[Ashwini]

yes U r correct!

[newms]

Getting a MS may be a disadvantage if it's a terminal degree without research experience, because some PhD adcomms might look at it as you not being sure about what you want to do. A research based MS wouldn't be a disadvantage and may give you an advantage if you're able to get good publications, but at the same time, if your app is strong enough to apply for a PhD then getting a research based MS first represents an opportunity cost, since when you go for a PhD it might cost you an extra year or two, than if you applied straight for PhD after finishing your BS. So if you want to go for a PhD eventually, then apply directly to the PhD or MS/PhD programs unless you feel like you need to get some publications for a stronger application, in which case a research based may improve your chances. Keep in mind also, that most top 20 programs don't expect applicants to have publications (although for top 5 programs you probably will need a publication).

As for your other questions:

1) From what I've seen most research based MS students do tend to get a publication of two out before they get their MS. Some don't though.

2) I'm not sure how long it takes journals to get published in CS. As you say, conferences are where most of the top CS papers get published and the top conferences have submission deadlines around 6 months or less before the conference. Why would you want to publish in a journal over a conference? You can write up a technical report if you want to write longer articles and refer to the report in your paper.

3) I don't know how common it is for MS semester projects to lead to publications, but it does happen and your goal, if you're doing a research based semester project, should really be to produce a publishable article based on your research.

Keep in mind that funding is scarce for MS students at most schools so unless you have a really good reason for going for a MS first, you should apply to PhD or MS/Phd programs if you want to eventually get a PhD (I'm speaking specifically of US schools here).

[President]

Thanks!

I'll definitely be shooting for a research M.S. The reason why, is because I will be satisfying pretty much the "minimum" degree requirements for a B.A. CS degree, since I decided "Oh! I'm gonna major in CS" during my junior year... I can't do any quality research when I barely know a thing about the field. The M.S. degree would require three semesters for me to complete, and I won't be attending a different school for it! I'm going to study with profs that I've spoken with at my current school (Top 50 CS department, and there are some pretty awesome-O faculty members here.. most interested in Comp Linguistics, NLP, Information Retrieval, and Databases).

So okay, I guess journal publication is actually useless; I'll just stick to conference papers... and I guess that'll give me more opportunities to resubmit for eventual publication on a different conference!

Thanks for help shedding some light on my situation. I can see that pursuing an M.S. isn't that much of a set back for me (besides the opportunity costs).

[BKMD]

A few thoughts. I think you're right that they will expect more of students with an MS than with a BS, and if you have publications as an undergrad, that's more impressive than as an MS, simply because it's less rare. But that doesn't mean that it will disadvantage you in any way to have a MS - adcomms are looking for candidates with the right background and experience, and an MS can only help you in this regard. The only way I think an MS could hurt is if you bomb it and get a low GPA or fail to impress the people who will eventually write your letters of recommendation. If you have the drive and you plan to do a PhD, then this scenario is unlikely.

As for publications - many but not all masters students will publish during this time, and it's definitely something to aim for if you want to get into a top school for a PhD. I know you said you started late in the game with CS so it may be hard to get into research right away, but if at all possible, I would start working on a publication as soon as you can. This is because it can take some time to get it published, and remember that you'll get applying for a PhD a year before you actually attend, which means that anything you publish in the last year of your MS will not appear in your PhD applications, so you need to do this stuff in your first year.

I wouldn't say that journal publications are pointless - it's just that in CS, conference pubs tend to get more exposure, and journals are generally just extended versions of existing conference papers (which is a good thing to have, but it's not what you should aim for as a starting point). It's also true that it can take a year or more to get it published, because the review process is slow and careful, and you will have to make many revisions, then those revisions are reviewed again, etc.. (Though sometimes journals periodically have "special issues" which usually have a rushed review process, so you might be able to get something finished within half a year.) I would recommend submitting to conferences, though you will have much higher impact if you submit to the top conferences in your field, and not just any conference or workshop.

It's not "hard" to get something published if it's a very good paper. If it's only a "good" paper, then you'll probably eventually get it accepted, but it can be harder - the thing is that many conferences have more good papers than they can accept, so it's kind of a crapshoot which borderline papers get accepted and which don't - it also depends on the reviewers you happen to get, and it's a fact of life that some reviewers won't understand what you're doing or won't put the effort into giving a thorough review, so plenty of good papers get mediocre marks. So my point is that if you start early, you can deal with paper rejections better because hopefully you'll still have time to revise the paper and resubmit to the next conference.

Good luck!

[President]

One last question guys:

I'm actually planning to take a year off after the MS program so that I can see my papers hit the proceedings when it's time for my applications to be sent in. If I'm gunning for a top 20 program, does it hurt to publish multiple articles at a low-ranking conference?

[Amogh]

One last question guys:

I'm actually planning to take a year off after the MS program so that I can see my papers hit the proceedings when it's time for my applications to be sent in. If I'm gunning for a top 20 program, does it hurt to publish multiple articles at a low-ranking conference?

Logically, it wouldnt hurt you but wouldn't help you either. You would be better served publishing one really awesome paper at a top conf than publishing multiple papers at low ranked ones

[President]

Thanks for all of your input guys! It really helped and shed some light on my situation. Another question:

When it comes to selecting departments I know that you should take a look at the faculty research profiles. If a school only has one or two professors conducting research on your field of interest, is it unwise to attend that school? Or is it still feasible to apply to that school with the intentions of working only with that faculty member (Let's assume there are at least two, since counting on just one professor is risky = sabbatical)?

I have a feeling I should only be applying to schools that have an established "research center" or "group" in my field of interest. Am I correct?

Thanks!

[newms]

I have a feeling I should only be applying to schools that have an established "research center" or "group" in my field of interest. Am I correct?

That would be ideal. When you say group, keep in mind that a 'group' can refer to a professor, his/her post doc(s) if any, and graduate students and sometimes one or two undergraduates. So as long as there is one prof that you are interested in working with, it is very likely that he or she would have their own group or lab. Ideally you also want to make sure that there is more than one professor doing research in your area that you could work with, since you never know what could happen in the future.

[BKMD]

It is nice if the school has a center in your area so that there are many people you can collaborate with. Realistically, however, most of your research will be done with one or two professors, so if you have to choose between a great fit with one or two professors or a good fit with several faculty, I would choose the former option. I do agree that it's risky to join a department if there is only one professor you think you could work with, since it's not that uncommon for professors to move between universities, or to leave academia for industry.

[President]

It is nice if the school has a center in your area so that there are many people you can collaborate with. Realistically, however, most of your research will be done with one or two professors, so if you have to choose between a great fit with one or two professors or a good fit with several faculty, I would choose the former option. I do agree that it's risky to join a department if there is only one professor you think you could work with, since it's not that uncommon for professors to move between universities, or to leave academia for industry.

That makes a lot of sense to me:D One of the reasons why I'll be gunning for a school with the latter option is because, at the CS department that I attend, there is only one professor who conducts research on NLP. The rest of the Comp Ling and NLP people reside in the Linguistics department. It's so frustrating, because my research interests are not exactly in line with his; great guy, though. Regardless, I take it that it's pretty common for a person to apply to a CS department with the intentions of working with just ONE of the faculty members at that department. Thanks guys.

[anacron]

, I take it that it's pretty common for a person to apply to a CS department with the intentions of working with just ONE of the faculty members at that department.

I'm not sure if this is actually the case...it may be beneficial to find who the potential advisor might be but its probably better to focus on the subject of interest.

For me at least, my current advisor is set to leave my university for a permanent position next Fall. Add to this the likelihood of senior faculty taking sabbaticals or spending more time managing the department etc. you may not have the same advisor that you started with or, at least, not with the level of involvement you expect.

If you find a school that is well known/established in your area of interest then they would be more likely to add faculty immediately in order to keep up the standards thus lessening the impact on dependant PhD students. So I wouldn't narrow down the departmental search to only one professor but rather focus on a research group (which is most often comprised of several professors in the same area).

I'm currently an undergrad so faculty changes affect me very little at this point, but some of the PhD students in my group now have to bring the new faculty member up to speed and hope he/she is interested in their project. Although this may not be a problem since they do screen the faculty candidates based on which department and corresponding research group opening the college is trying to fill.

This was most of the information/advise my advisor gave me, hope it helps.

On a related note, do adcoms differentiate between BS and MS students when they review applications? or review them separately and have allotments on how many MS vs BS applicants they will take?