Notes

• Advice from my advisor
• Start writing papers even before start working on it
  • the coding takes time, the execution takes more time
  • but by writing you can find the caveat faster
• Paper (when finished) must have only ONE SHARP idea
  • if you have multiple ideas, write multiple papers
• Write paper around one single idea
  • maybe when you are writing you dont know which idea
  • but when you are at the end, you should know which idea
  • and then reframe the whole paper circling around that idea
• Think about your readers
  • you want to give reader knowledge in the easiest way possible
  • read from the reader perspective
    • is it easy to understand?
    • is you main idea just ping on the user's mind?
• send to your colleagues
  • ask them where they got lost
  • or where they got bored
  • or where they lose the enthusiasm to read it
• Introduction
  • write motivation
  • write contribution
    • write contributions as interesting that the readers say "oh gosh, we need to read the paper to find it out how they did it"
  • use tension-release cycle -- there is a problem (tension), but we solved (release) it by ...
  • forward reference your sections from claims where the claim was substantiated
• Related work
  • go to the end just before conclusion
  • its not interesting, only used to connect your papers to other studies
  • but only read by people who is finding papers
  • you dont have to make others look bad to be good
  • give credit where its due and show what you have improved on that
• Summarized advices
  • dont wait: start writing, even with not fully blown idea
  • identify your key idea (at least by the end of writing)
  • tell a story
    • Pose the problem, ask a question, pose a solution, note problems that arise, address them, denouement
  • nail your contributions (ideally in 1st page / intro)
  • related work: later
  • Put your readers first (use examples that will make easier for them to understand)
  • Send to colleagues for review
• plots
  • Best Color Palettes for Scientific Figures and Data Visualizations

References:

• [How to Write a Great Research Paper - Simon Peyton Jones](https://simon.peytonjones.org/great-research-paper/
• https://www.alignmentforum.org/posts/eJGptPbbFPZGLpjsp/highly-opinionated-advice-on-how-to-write-ml-papers
• Figure Creation Tutorial: Making a Figure 1 - Maxwell Forbes
• yzhao062/cs-paper-checklist: A final sanity checklist to help your CS paper get accepted, not desk rejected.
• writing advice from my advisor
• [Paper writing checklist](web.ece.ucsb.edu/~zhengzhang/paper writing checklist_v2.pdf)
• A must read: https://maxwellforbes.com/posts/how-to-get-a-paper-accepted
• https://www.youtube.com/watch?v=imEtTnQKt4M
• https://www.cs.ryerson.ca/~wangcs/resources/How-to-write-a-good-CVPR-submission.pdf
• https://deviparikh.com/citizenofcvpr/static/slides/malik_write_good_paper.pdf
• [ ]
• dgp.toronto.edu/~hertzman/courses/gradSkills/2010/writing.pdf
• 08 How to write papers so that people can read them
• LEADERSHIP LAB: The Craft of Writing Effectively
• The PhD Metagame: How to Get Your Paper Accepted - Maxwell Forbes
• Write the Paper First
• Tips for Writing Technical Papers
• dgp.toronto.edu/~hertzman/courses/gradSkills/2010/writing.pdf
• (669) 08 How to write papers so that people can read them - YouTube

Slides

• deviparikh.com/citizenofcvpr/static/slides/malik_write_good_paper.pdf
• [ ]

Paper review or rebuttal process

• https://docs.google.com/document/d/1iNmLWb0z8F4Td-WykXdhTQfzsX26zXm7pY2e34SVmFo/edit?tab=t.0
• https://www.jakobfoerster.com/how-to-rebuttal-ml-paper

Some better resources:

• Advice for Research Students (and others)
• Research skills: resources - Simon Peyton Jones
• [people.csail.mit.edu/billf/talks/10minFreeman2013.pdf](https://people.csail.mit.edu/billf/talks/10minFreeman2013.pdf
• https://www.youtube.com/watch?v=WQF_m4kD32w
• https://www.cs.toronto.edu/~miller/Research/writing.html
• people.csail.mit.edu/billf/publications/How_To_Do_Research.pdf
• https://github.com/Niko-Group/paper_writing_info/tree/master