美国教授如何判别学生是否有潜力美国教授如何判别学生是否有潜力美国教授如何判别学生是否有潜力很久以前的文章，不过有意义。作者: Sanjoy Das, Assistant Professor, Kansas State University Ph.D. 1994, Louisiana State University. Evolutionary algorithms, neural networks, artificial intelligence, discrete optimization. International applicants seeking assistantship/admission: please do not send me emails blindly without reading this Every day I receive emails from international candidates for information, for advice, and especially requesting financial assistance. Unfortunately, I delete most of them without even reading them carefully. I get so many irrelevant emails that cannot reply to each. Extremely rarely has one caught my attention. Please do not send me emails without reading this first. I am only interested in corresponding to candidates who are exceptional, and have a strong interest and experience in my immediate area of specialization, and I highly prefer ones with a Masters degree. Here are two sample letters: A sample letter that will never get the candidate an assistantship with me: Unwanted email Respected sir, I am very interested in going to the US for higher studies. I am seeking admission and financial assistance in your esteemed University. I am very interested in artificial intelligence and computer architecture. I have programming experience in C+, C and . I am also familiar with several software packages in the Windows environment (MS Word, Excel Spreadsheet etc.). I also have networking certification from NOVELL. My GRE score is 2050, and I got 99 percentile in the TOEFL exam. I request you to spare a few moments from your precious time to go through my attached resume. Sir - if you are kind enough to offer me assistantship, I promise to work very hard for you. I have a very good academic record since primary school. Alternately, can you tell me what are my chances of getting assistantship from your department? Can you kindly inform me about the last dates for Summer term 2002 before which I can apply? Thank you for your valuable time, Yours obediently ABC My comments: I simply delete these emails, which I consider garbage . The person seems to be too obsequious and flattering. In the emails that I do read (occasionally), I do a first scan for keywords like “precious“, “respected“ etc. deleting the email if I find one. Obviously the person is not professional enough. The person is interested in artificial intelligence and computer architecture - these are quite unrelated areas, making me think that he is trying to cover as many professors as possible to get aid from any possible source. Besides, he does not have anything in support of his claim (about his interest in artificial intelligence). He emphasizes on his software skills, making me think that his goal is to get a programming job in the US and higher studies is only to allow him access to the US job market. His GRE scores are OK, but not exceptional. He asks me to tell him what his chances of getting an assistantship are, elsewhere in the department. I have absolutely no way of finding that out, and even if I knew, I would not bother to reply back. He also asks me to give him information about summer deadlines. How am I supposed to know that? (I would typically expect that this information should be available from the University web site, or from the grad school, but my guess is as good as yours.) Finally, this letter is too generic . He probably uses exactly the same format for hundreds of email letters to various professors throughout the country, leaving no stone uncovered. I receive hundreds of email like this. Regretfully, I cannot answer everyone, and these applications do not interest me at all. Now, here is a letter that would DEFINITELY attract my attention: Excellent email Dear Prof. Das, I am applying to KSU (amongst other places) for a Ph.D., and I am very interested in your research areas. I would like to explore the possibility of doing my research with you, with financial support. One possible area that would interest me a lot is in applying Tikhonov regularization to your problem in computational neuroscience, since I understand that neural data can be very noisy, and yet sparse. Alternately, I am also interested in applying evolutionary algorithms to your genetic neural network parameter estimation. One of my other interests in in parallel processing. Are you considering parallel implementations for any of your optimization/estimation algorithms? Considering that the projects that you list in your web site would probably be very computationally intensive, that may be a worthwhile strategy to pursue. I presented a paper entitled “Hybrid Gradient Descent Based Training of Probabilistic Recurrent Networks“ in the International Conference on Neural Networks and Image Processing, Bangalore, last year. It is a reviewed publication. If have attached a copy of the article in pdf format. If you have available funds, and are interested in my application please let me know soon. XYZ