ACOUST%BNR.CA@MITVMA.MIT.EDU ("Nahum Goldmann", N.) (09/01/88)
Date: Tue, 30 Aug 88 10:25 EDT To: Nick Papadakis <AIList@AI.AI.MIT.EDU> From: "Nahum (N.) Goldmann" <ACOUST%BNR.CA@MITVMA.MIT.EDU> Sender: "Nahum (N.) Goldmann" <ACOUST%BNR.CA@MITVMA.MIT.EDU> Subject: Where should she go? Peter Webb writes (AILIst, v8, #60) > A friend of mine wants to get her PhD in Computer Science, > specializing in the Machine Learning aspect of Artificial Intelligence. > She has been to the library and collected a list of likely schools, but > the list is too long for her to apply to all the schools on it. > Accordingly, she asked me if I would ask the net for suggestions. If > you wanted to study machine learning, where would you go and why? I believe that the procedure of evaluating a SCHOOL is only suitable when the undergraduate education is being considered. For a graduate student, especially a PhD candidate, the first step to prove his or her scientific maturity is to identify the area of his/her OWN interest. Myself, I am expecting a PhD student to come with a reasonable degree of aggression and violence to prove that: a) I don't understand anything in my own area of expertise; b) It really does not matter, since my area is doomed in any case; c) The applicant has a marvellous idea which will save both me and the mankind (the humankind?) from the absolete approach (not that it will happen at the end of the exersize, but it is a different story). Would anything less do? On a practical note, I'd advise her to do the following: 1) Find a couple of good reviews in the library which deal with the subject (machine learning?). 2) Loosely identify 2-3 sub-areas of interest. 3) Find fresh publications in these areas. Based on them, define which circle of problems/methods really appeal to her. Just intuition will do. 4) Based on the same publications identify major players in these areas whose works sound exciting. 5) Contact these INDIVIDUALS and ask their advice (who is the best PERSON to do YOUR research with). You'll be surprised how much more informative their responses will be than what you will get "at random". Yes, they may not speak about dormitories and the "perceived importance" of the college, but does it really matter where a good researcher is located? What if it is in Australia, Japan, or the UK? For a PhD student it should not be a major obstacle. 6) Come to the person selected and convince him/her that without you (see a-c above...). Propose a couple of research subjects. At the end, settle for the subject HE gives to you). It is less likely that she'll have a major dissapointment at the end of this long exersize. Sorry for the basic staff. It's just that I've seen so many PhD's who would be far happier if they were insurance agents... If only someone would explain the basic facts of scientific life to them beforehand... Good luck to your friend (at least she asked the question)! Greetings and love Nahum Goldmann (613)763-2329 e-mail <acoust@bnr>