1 / 33

Survival skills for students (an empirically-biased perspective)

Survival skills for students (an empirically-biased perspective). Joe Beck. THINK ABOUT HOW TO PEEL OUT GRAD LESSONS FOR UNDERGRADS WHERE RELEVANT. Abc. Sources. Graduate student professional development series – Umass

herve
Download Presentation

Survival skills for students (an empirically-biased perspective)

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Survival skills for students(an empirically-biased perspective) Joe Beck

  2. THINK ABOUT HOW TO PEEL OUT GRAD LESSONS FOR UNDERGRADS WHERE RELEVANT Abc

  3. Sources • Graduate student professional development series – Umass • Talk on “How to be a research advisor” by Manuela Veloso (seemed more geared towards students!) • Gave me direct and indirect lessons • Personal observations

  4. First things first… • If you’re here without funding and a research area, need to address that first • (I also went to grad school without funding) • Getting “A” grades in your courses is a good start • But not nearly enough • 1. Too slow – building a track record takes time • 2. Want more than good course grades

  5. Financial facts of life • Grad students are expensive! • About $55,000 per year (here – a bit less at UMass, much more at CMU) • (salary + tuition + overhead) • Grants have specific objectives • Not a casual decision to hire someone as a Research Assistant

  6. How to impress potential advisors • Do an independent study • Be engaged • Make progress • Come up with ideas the faculty member didn’t • But it’s (maybe) too late to sign up! • Doesn’t matter, can be informal • Read some papers in his area and ask smart questions • Go to lab meetings and ask smart questions

  7. What not do • Write an email: “Dear Professor, I have a very strong student and graduated with a 3.94 GPA from…” • Probably stopped reading then • A targeted email: “Dear Professor Smith, I liked your approach of using genetic algorithms to find a good used car. I’d like to stop by and…” • More likely • But higher likelihood is catch Smith in his office between meetings

  8. Teaching Assistantship • Department will look to recruit TAs (not high turnover, but usually a few throughout the year) • Will look for people who stand out in courses • Do well • Ask and answer questions – communication skills • (don’t go overboard)

  9. Keeping you and your (research) advisor happy • Be proactive • Great advice from Manuela Veloso: go to a meeting with your advisor with an agenda of what you will work on next • Shows advisor you’ve thought about it • Easier for advisor to just say “yes” • If you come with no agenda, advisor will come up with one (we’re good at it) – might not be what you’re interested in

  10. Research advisor: most important person for a PhD student • Not someone you want to irritate • Triggers will vary from person to person • Typically not fired for making mistakes • Not getting things done is much worse • (make sure you have a list of your weekly accomplishments)

  11. You and your advisor, getting along • Advisor is not someone you want to irritate • Missing meetings • Not being prepared for meetings • Not really trying • If you cannot get along with your advisor, consider other advisors (particularly if PhD student)

  12. Undergraduates • A lot of what I just said applies to you too • How to be productive and get along is fairly consistent • Getting noticed is major difference • Partially driven by finances • Also in what you have to offer

  13. Working with a research lab • Faculty are much more interested in you than you might expect • You don’t know much – yet • Big part of knowledge is understanding how things work in the lab (software, people, process) • Takes time to learn – not in classes •  Freshmen are generally of interest • (Sonia: should have some course background as foundation)

  14. Other reasons to work in a lab • Learn a lot about process • Useful experience for resume • Letters of rec • If you have work study, is a no brainer • (wish someone had told me)

  15. How? • “90% of success is showing up.” – Woody Allen • Work study fair • Check out list of faculty by research area • Wander into a lab that sounds interesting and start asking questions • Volunteer to help out

  16. Essential skills • Programming • Data manipulation • Writing documents

  17. Programming • Get good at a language • Don’t care which (within reason) • Ability to think in a language lets you be much more productive • If your group has standardized around a technology, may have to use that • Otherwise, choice is not that important

  18. Data manipulation (Xiaolu) • Empirical science involves data • Often not in quite the format you want • Want to slice it differently • Or compute different features • Or merge two data sets… • Get good at this part of your job • Tokenizing strings, standard ways of computing things, when to use Excel

  19. Learn to use a spreadsheet • Surprisingly versatile! • Code a simulation • Graph data • Analyze data • Compute features • Do analyses I do not know how to do in other packages (e.g., how does the correlation change based on the amount of data)

  20. Writing: technical aspects • Learn how to use a word processor • Don’t care which • But your life will be easier if you pick MS Word (Joe) • LaTex makes sense in some communities • What do I mean by use? • How to use style templates • How to insert captions and cross references • Headers and footers • WPI has classes on these things

  21. Understanding what you read (Ben, Dmitry) • If don’t understand a paper, and it’s important • Try related references • Contacting the author • Asking other members of the lab • Focus on… • Abstract (see if it’s important) • intro, conclusionand headings • figures and tables contain a lot of empirical results • If really useful, or part of your lab, might be worth reading all of it • Read in layers • Quick pass to see what they’re doing • Second pass, do some quick checking of references • Third pass, goal is to really understand the paper

  22. Finding a research problem • A good topic should be one that others care about and that you can accomplish

  23. Problem • If others care about it, odds are others have looked at the issue, or are studying it now • Why will you succeed?

  24. Because I’m smart! • Yes, you are • But so are most of the other folks working on this problem • But I’m smarter! • No, you’re not

  25. I’ll work really hard • Effort really does matter • But most people work on problems they like • And so work hard on it • Unless you have some combination of insomnia, perseverance, and no social life, unlikely you’ll be an outlier

  26. You need a “secret weapon” (Herb Simon) • Why will you succeed instead of others? • Need something others don’t have • A mental model • A piece of equipment • A data source • Specific training • A bigger budget

  27. Picking a good question • Should not care about answer • Should be motivated • How to turn horse race into something more interesting

  28. A horse race? • One technique will do better • There are two techniques X and Y, let’s see which one works better… • Who cares? Inventors of X and Y. Anyone who is looking at using X and Y in very similar circumstances

  29. An example • Imagine we want to create a new robot to handle navigation in a complex environment • What are factors that might influence difficulty?

  30. What are some things that could influence performance? • Size of maze • Reflectivity of surface • Tractiomn • Budget for robot • Types of sensors • Gps available?

  31. Contrast two papers • We developed a new robot for navigation task ABC and it did 7% better than the prior best. • We developed a new robot for navigation task ABC, and hypothesized that it’s better sensors would enable it to do better in tasks in non-cramped environments. We found that…

  32. Take a deep breath • Plenty of people before you have gone to graduate school • Many of them finished • And some of them were idiots • So your chances are better than you think :-)

  33. http://www.wpi.edu/academics/cs/research-groups.html

More Related