Scientific writing tips for ML researchers*12p9A2_w6aJaTJZzJYdpDg.png

Hot “call for papers” season is right around the corner. Here is a list of scientific writing tips for all you ML researchers.

  • An abstract is not for venting…
  • The abstract: the problem, state-of-the-art, contributions, and results
  • Avoid teasers, be more explicit
  • Avoid generic statements…
  • Aim to make a first impression with interesting observations
  • Problem statement, then solution, followed by results
  • Real world use cases are juicy and motivating
  • Avoid digressions, especially negative ones
  • Neatly organize your paper: sections, subsections, bullets, etc.
  • Figures tell a story, no-figures should have the same effect
  • Basics are boring, get to the meat of the matter (contributions)
  • Anticipate questions, and answer them as best as you can
  • It’s all about “we” not “I”
  • Sentences in isolation should be “unambiguous” and “objective”
  • Avoid boasting, if you are not 100% sure of your claims
  • An opinion should start as “in our opinion,…”
  • Long sentences usually spell disaster…
  • Keep sentences as compact and clear as possible…
  • Sentences should have a flow, like a song
  • Intensifiers only weaken your argument
  • Pay att. to relationship between your subjects, verbs and modifiers
  • Don’t cite for sake of citing…only cite the most relevant works
  • For readability, let citations roam freely all over your paper
  • Cite, cite, cite, cite, and keep on citing for as much as is allowed

A checklist adopted from “Heuristics for Scientific Writing (a Machine Learning Perspective) — @zacharylipton

Made with ? by @omarsar0

? Find checklist version here

Scientific writing tips for ML researchers was originally published in DAIR on Medium, where people are continuing the conversation by highlighting and responding to this story.