BYR Achieve · 镜像论坛

原帖地址： http://www.ece.cmu.edu/~koopman/student_info/prospective.html ____________________________________________________________ Dear Prospective Student, Presumably you're reading this because you would like to be a graduate student, possibly at Carnegie Mellon University, and possibly working with me. This web page is an attempt to give you more information while managing my e-mail load so I can spend more time on my current students, teaching, and various obligations. Unfortunately, I am far too overloaded to respond to individual queries regarding admission. None of this is official policy of anyone, and much of it varies from professor to professor and from school to school. The below are in Q&A format in the tradition of FAQs: Will sending you e-mail increase my chance of admission? In most cases not. I get hundreds of e-mails each year from prospective students. Unfortunately, most of them are essentially spamming attacks on all university professors in my field or in my department. It's difficult for me to sort it all out, and spending time on that takes time away from my current students who legitimately deserve my primary attention. The other professors I have asked completely ignore such e-mail. I can't afford to spend time responding to spam, and have few efficient ways of knowing what's spam vs. what are well-considered, individually composed, earnest queries. In every case I've seen, spending time making your application look good is more productive than e-mail campaigns. Sending me e-mail that includes a resume and a lot of other information is a waste of effort because I simply don't have time to read all the resumes I get sent via e-mail -- put that effort into your official grad. school application. Who decides who gets admitted to be your student? For ECE admissions I personally make my own admission decisions. There is no committee deciding for me who I get to admit (although I have to have a good story to tell the department if GRE/TOEFL scores are way below average for our students). I see summary information on every application sent to the department with "Computer Engineering" checked as the first choice area, and depending on how many open positions I have and the number of applications submitted, I fully read dozens or sometimes hundreds of folders when making admission decisions. For the ISRI program we use a committee approach in which several of us read each folder, and pretty much every professor who is a potential matchup reads the folders of admitted students. It all takes a lot of time. How do you pick students? Ultimately it is my personal judgement call on whether you will succeed at CMU, and whether you will be interested in working on the research topics I am planning to work on. Additionally I am limited in the number of students I physically have time to supervise and the number of students I can fund. Every year there are very good students that I can't admit simply because I have limited time, facilities, and support money. In the coming several years I expect to admit about one or two students per year to my group out of the many hundreds of applications we receive for the computer engineering area. I only admit PhD students. I can't pay for graduate school; what are the chances of assistance? Your chances are 100% if you are accepted for the PhD program. I fully fund all of my students, including tuition and a stipend that nicely covers living expenses in Pittsburgh (pretty much the same stipend as other big name schools, but far lower cost of living). Typically students get a research assistanceship, and some get a full or partial fellowship. It is typical for your admission letter to tell you which kind of funding you will receive and guarantee it for one or two years contingent on satisfactory progress. As a practical matter, it would be an extremely unusual situation for you to ever be asked to pay tuition while you are here -- I haven't ever done that. If we don't think enough of you to pay your tuition, then you probably shouldn't be in the PhD program. Coming with your own fellowship funding is nice, and gives increased flexibility on the projects we can work on, but doesn't affect your chance of admission directly because I select based on abilities and match of interests rather than on funding. Complicated letters and bank statements proving you can pay your own way are completely ignored -- don't waste your time on them in ECE or ISRI even if the CMU official policy is to send them for other programs NOTE: the 100% chance of funding does not apply to MS-only programs. I don't admit students to those programs since they are based mostly on coursework and not a research-based degree. Be sure that you understand the rules for the particular program you're applying for. For a Ph.D. student in ECE or a Ph.D. student in ISRI, you can anticipate full support per the above. I got a miserable GRE score because I was sick that day, etc.; what should I do? Your best bet is to take GRE's early in the season and take them again if you have a bad day and get a bad score. I'd love to have a better assessment mechanism than the GRE. But the fact is we have to have some calibrated measurement source and that is the one that is available to us. The applicants I admit are usually near or above the 90th percentile on all three GRE scores. (I personally consider verbal skills to be important, because a major product of research groups is written material. If you can't communicate what you have learned, then there isn't much point doing research.) Beyond the GRE scores, we also look at your undergrad GPA, recommendation letters, and essay. High GRE scores by themselves won't get you in, but low scores will be a problem. I note that there have been some apparent problems with the integrity of GRE scores, especially outside the US. This means I must put more emphasis on things like recommendation letters from people and schools whose reputation I know of. How many of your students actually graduate? There are department policies in place to make sure every student gets more than a fair chance to get a degree. We have a low "drop-out" rate, and a high rate of doctoral students passing the qualifiers. Once you come we expect you to succeed; there is no artificial quota for washing out or otherwise culling students. Every step is one where you succeed on your own merits rather than as part of a competition. In general about 1 in 8 don't pass the qualifiers even after two tries, and another 1 in 8 don't make it all the way to graduation for a wide variety of reasons. Thus, probably 3 in 4 who want a Ph.D. achieve that. My attrition rate is about average or slightly better. The numbers don't tell the whole story -- really it is all about whether you have what it takes. How long does it take to get a degree? Ph.D. student time varies all over the place, from a record of four semesters after M.S. to almost eleven years, although I don't ever want to have one of my students take ten-plus years to graduate! 5 years past B.S. is a common goal to set and is achievable for most students if they remain focussed on their work. My students tend to graduate in average or slightly faster than average time. Further details are better discussed once you've been admitted and are deciding whether to come here. What research areas will we work on together? You should realize that when I admit I am often staffing a new research project that doesn't even appear on my web page. That being said, look at my research interests and current projects to get a flavor of what I do. You will be working on projects that are the similar or in new (but probably related) directions. My particular interests are in embedded systems, embedded communication networks, and system-level dependability. If you are admitted the department and/or I will help pay for a visit here so we can meet and talk about what you'd be doing in more detail before you have to decide on a school. The graduate research experience here is very personal, and involves significant amounts of one-on-one faculty/grad. student interaction. What about cost of living and housing? This is a key question you should ask of any university to which you apply. Pittsburgh is a very affordable city, and our grad. student stipend is roughly the same as the other first-tier schools. So, you can actually live on your stipend without a problem. The CMU housing office has a searchable rental database to give you an idea of housing costs at http://www.housing.cmu.edu/CommunityHousing/ . I suggest looking up a small apartment in Squirrel Hill at a distance of 1 to 2 miles from campus, which is where I live. Stipends vary a little depending on your status in our program, but for the 2003/2004 academic year entering grad students were given $1725/month cash salary in addition to tuition. To see what that translates into compared to other cities, use a cost-of-living salary calculator such as at http://www.homefair.com/homefair/cmr/salcalc.html (when I checked it told me that our $1925 stipend is roughly comparable to a $2839 monthly stipend in Palo Alto, CA in terms of buying power, although of course this is just a broad approximation). The numbers are probably higher now; I don't update this page that frequently. I'm not a U.S. citizen -- how does that affect my chances? Unfortunately, if you are not a U.S. citizen and you don't hold a green card, your chances for admission are very significantly reduced. One simple reason is that we get an overwhelming number of applications from overseas. The extra work and time we spend on cultural acclimation and language skills for these students means that we can only accept a few of them, and of course this is a very few out of a huge application pool. If you already have a degree from a U.S. institution, that helps considerably with this issue. Also, it is more difficult to obtain funding for non-U.S. students, and so further increases the effort required to bring them here. Historically about half of our department's graduate students are non-U.S., although that proportion tends to be closer to one quarter in my research group. So, is there any way to actually improve my chances of admission? Mere admission shouldn't be your goal. Your goal should be finding a professor to work with who has compatible interests and making sure you communicate those interests and your special qualifications for that work. If you want to get my attention, the best way to do that is to demonstrate you have a combination of the following (simply having high scores doesn't indicate that you'll be happy here, and if you're not happy there's really no point in coming is there?). Excellent math and analytic skills (GRE scores, publications, won the state science fair in high school, etc.) Above average to excellent English skills, spoken & written (GRE scores, publications, "A" in technical writing course, debating team, student newspaper, Toastmaster's Club, etc.). Interest in the work I'm doing. Go ahead, make an insightful comment (or, even better, a valid criticism) about my research -- that will get my attention! When you apply it is very important to fill out the profile sheet with your official application and give priority to such areas as dependability/fault tolerance and embedded systems. It is also fine to list my name or the names of professors you have a particular interest in right on the "areas of interest" part of the form. Interest in doing research rather than simply getting a quick ticket to a higher paycheck. I'm going to invest considerable time one-on-one teaching you research skills and advanced problem-solving skills I learned both in universities and in industry. That takes time -- we don't just give you a handful of courses and send you on your way. Other evidence that you finish what you start (e.g., Eagle Scout, 4-H Teen Leader, athletics, worked your way through school, or almost anything substantive). Industry or military experience is an especially good thing to have, but not required. If you can't already speak English with native or nearly native fluency, your probability of admission is low. If you are non-native speaker, the best way to provide evidence of fluency is to have lived in an English-speaking country for a few years, preferably while getting a degree. (Or, come from a country that historically produces excellent English speakers.) If you really are fluent but it isn't obvious from your credentials, I can arrange for you to get an oral interview/assessment with one of our international student counselors via telephone if you are one of the final few I am considering (but please, please don't waste either of our time if you just went to a TOEFL cram school -- that isn't what I'm talking about here). Once I apply, how can I find out my chances for being admitted? In the next few years I expect to keep my student group relatively small. Probably I will admit one student each Fall. Competition for entrance to Carnegie Mellon is intense, and there are many more qualified applicants than there are positions available. Overall, only a few percent of applicants are admitted. Unfortunately, we can't possibly give everyone who applies a status of their chances beyond whether we got material that was submitted. And, it is impossible for me to look at information you e-mail to me and give you an estimate of your chance. If you have read the above and understood it, you will be able to tell if you have a chance or not. Beyond that, there are too many variables, and too many constraints. Even a late decision on your application isn't conclusive, because sometimes an internal financial issue must be cleared up before we can make an offer. So, if you ask me questions such as "how am I doing," or "when will I hear an answer," or "here is my information, do I have a chance," I simply can't answer them and will not reply. Keep in mind that only a few percent of applicants are admitted, and many highly qualified candidates are not accepted for a variety of reasons having to do with funding and capacity constraints. So there is no shame in not being admitted. Beyond that, it is not wise to make your application to Carnegie Mellon the only one you submit. But, of course, good students do get in every year, and you could be one of them. One thing that you can be sure of is that if you are admitted, we'll make 100% sure that you are notified immediately (I make that phone call myself). OK, I'm going to give it a shot; how do I apply? See http://www.ece.cmu.edu/ for admission forms. Unfortunately, my e-mail load is just too heavy to be able to respond to queries from students. I take great pains to go through the admission packages submitted. If you indicate that you want to do "embedded" or "dependable" computing in your application, I'll see your application. Similarly, you might want to mention the names of various professors you'd like to work with in your application essay (that is a very effective technique to make sure the right professor actually sees your application). With best regards, Phil Koopman Associate Professor Carnegie Mellon University ECE Department & Institute for Software Research International

转一篇CMU教授写的选人标准～