$30 off During Our Annual Pro Sale. View Details »

Advice to New Researchers: Simplify Research

Advice to New Researchers: Simplify Research

Some tricks that worked for me:
* Break it down!
* Test Driven Development
* Metrics
* Iterate / Incremental Progress
* Be happy: Tangible rewards

Arnab Nandi

June 24, 2014
Tweet

More Decks by Arnab Nandi

Other Decks in Education

Transcript

  1. simplifying research

    Arnab Nandi

    Assistant Professor!
    Computer Science & Engineering
    The Ohio State University
    interactive data systems
    research group at ohio state

    View Slide

  2. 2012
    Hi! I’m Arnab.
    interactive data systems
    research group at ohio state

    GestureDB  
    Querying  beyond  Keyboards  
     
    DICE:  Interac9ve  Analy9cs  
    Ad-­‐hoc  Data  Analy9cs  &  Explora9on  

    View Slide

  3. Don’t let it scare you
    RESEARCH

    View Slide

  4. Problem"
    HARD
    RESEARCH

    Process"
    SIMPLE

    View Slide

  5. tricks that worked for me

    View Slide

  6. View Slide

  7. Test-Driven Development
    •  Unit tests
    •  1 test per “thing”
    •  Top-down
    •  If Research Problem was solved,
    what would be possible?
    •  Bottom-up
    •  Naïve algorithm
    •  Strawman
    •  Free progress bar for research!

    View Slide

  8. Metrics, Metrics, Metrics
    •  Really easy to get distracted
    •  Anecdote: sub-second threshold in DICE

    View Slide

  9. Incremental Progress

    View Slide

  10. Incremental Progress
    •  Journal your research!
    •  Daily TODO.txt
    •  Weekly Snippets to advisor
    •  Streak Calendar

    View Slide

  11. Iterate, Iterate, Iterate
    You are not
    going to get it
    right the first
    time


    (if you do, let’s chat.)

    View Slide

  12. Don’t become a zombie
    •  Stagnation, Zero excitement, Feeling Lost
    •  Internships
    •  Take breaks (real vacations)
    •  Don’t let your ego get in the way
    •  It’s ok to quit a problem
    •  (instead of the PhD!)
    •  Breadth is a strength
    PhD…  

    View Slide

  13. Be happy: Tangible rewards
    •  Challenge: Research problems, submissions are “abstract” efforts
    •  Idea: Align “abstract” efforts with tangible rewards
    •  Example: Paper Submission: Hardware Upgrade
    •  Note: not publication, submission
    •  Important to pick the right metrics here

    •  Remove resource constraints ASAP
    •  Books, furniture, hard drives, food
    •  Worth it even if your univ. doesn’t reimburse!
    •  “Grand Slam”

    View Slide

  14. Summary: Simplify Research!

    •  Break it down!
    •  Test Driven Development
    •  Metrics
    •  Iterate / Incremental Progress
    •  Be happy: Tangible rewards

    View Slide

  15. Thanks!
    interactive data systems
    research group at ohio state
    papers, videos and more at http://interact.osu.edu | http://arnab.org

    View Slide