Literature Study Tips
An alternate resource for reading papers can also be found here
Table of Contents
Reading Research Papers
These are the notes from the amazing lecture by Andrew Ng, full video can be viewed here
Steps to follow
Organizing the papers
- Compile a list of resources
- Papers from arxiv and conferences
- Journals
- Medium/Github posts
- Articles/Blogs
- Skim through the list of resources
- Make a table, where rows represent lists of papers and a column for a metric of how much you understood through skimming (10-100%) - Now read the one with the lowest value of the metric, try to understand the paper, if you can’t, go to the references, and read those till you get a basic idea of the paper
- Keep doing this till you have a basic knowledge of the papers
- Then select the papers you feel is worthy enough to be completely read
- Reading around 5-20 papers, you’ll have some basic idea of the field and for implementing the works
- Reading around 50-100 papers, you’ll have a deep understanding to do in-depth research (it does not mean you have mastered the field :smile:)
Guide for a single paper
Take multiple passes through the papers
- First Pass
- Read Titles, Abstract, and Figures (only figures can sometimes summarize the entire paper)
- Second Pass
- Read more carefully the Introduction, Conclusion, Figures and then skim through the rest (skip related work if you’re not familiar with it in the second pass)
- Third Pass
- Read but skip or skim the Math
- Fourth Pass
- Whole thing but skip parts that don’t make sense
Questions to keep in mind
Ask yourself these questions while reading the paper
- What are the authors trying to accomplish in this work?
- What were the key elements of the approach in this work?
- What can you use yourself?
- What other references do you want to follow?
Some Sources of Papers
- Top Tier Conferences like NeurIPS, ICLR, CVPR, ICRA, RSS, IROS (more on this here)
- Subreddits
- Paper Reading Groups, Communities and Friends
General Tips
Math
To understand the Maths behind the paper
- Read a few passes and make detailed notes
- Try to rederive the math from scratch on blank paper
- If you can do this, then you can learn to derive your own novel algorithms - Eg. People from the art community sit in the art museum and they copy the work of the masters
Code
Download and run the open-source code and try to reimplement it from scratch
Long term advice
- Keep reading papers consistently
- In doing so you won’t gain expertise in one day, you won’t get a lot of knowledge from reading one paper a weekend. But if you keep doing this for a year, you’ll reach somewhere
- One great project is better than many lame projects
- Focus on the team (people you interact with)
- Maintain a work-life balance