Reading research papers can be hard and therefore I looked for strategies to optimize my paper readings. How should you read a research paper? How many papers should you read in order to understand a field broadly? There is a good talk on this topic from Andrew Ng, an Adjunct Professor at Stanford University.
Here are my notes on how to read research papers:
- Read 5-20 papers to get a basic understanding of a field
- Read 50 - 100 papers and you have a very good understanding
Steps to understand a field
Andrew Ng proposes the following three steps to understand a field better:
- Compile a list of papers (research papers, blog posts, …)
- Skip around the list
- Take mulitple passes through the papers
Make a list of 5 papers, read 10% of every paper and try to understand it. Try to find relevant other papers and update the list.
Steps to read a paper
- Read the Title, Abstract, Figures
- Read Introduction, Conclusions, Figures, skip rest (skim related work)
- Read the paper skip/skim the math
- Read whole paper but skip things that don’t make sense
Ask the following questions while reading
- What do the authors try to accomplish?
- What are the key elements of the approach?
- What can you use yourself?
- What other references do your want to follow?
Sources of papers
- Machine Learning Subreddit
- Conferences NIPS/ICML/ICLR
- Friends
My field of study is currently Intrusion Detectio (Cybersecurity) and Artificial Intelligence methods. You may think this is quite broad, and I agree. I try to narrow it down to a more specific topic. If you are interesting in my work just contact me and I’ll be happy to share experiences.