 |
|
|
|
 |
|
||
 |
 |
 |
|
 |
|
||
 |
|
||
 |
|
||
 |
|
||
 |
|
||
 |
|
||
 |
|
||
 |
|
||
 |
|
||
 |
|
||
 |
|
||
 |
|
||
 |
|
||
Reading papers is extremely important throughout the grad school, and should be especially intensive in the beginning. While books can give you a good background in some area, research papers will give you the necessary feeling of the open problems and the existing approaches to solving them. It is virtually impossible to have a solid good idea without seeing what else is there, and what people have already tried. So plan paper-reading in your schedule. Ideally, reading 1-2 papers on the average in a week should keep you well informed.
Some people have found it useful to print out the paper they are reading and mark it, underline important stuff, write notes on the margins and file it into a dedicated folder. Other people keep computer files with notes on papers they read. I keep on wishing to do one or both of these things but still haven't found time. However, both of these techniques seem like really good ideas and should simplify your life later, when you need to locate some relevant paper for a 'Related work' section.
Sometimes, if you are reading papers in a completely new field, they may seem overwhelming - crowded with formulas and jargon you don't understand. It usually helps if you don't read a paper from a beginning to an end, lingering on every formula, graph or proof. Try instead to read through it once superficially, looking to grasp the idea and important sections. Later, come back to those parts you found interesting and focus on them, trying to learn more. You can repeat this process many times. You may even leave some parts of the paper unresolved, if you think that these are not relevant for your research. Later, if they become relevant, you can always go back and re-read them. I call this technique 'multiple pass reading', since it resembles the philosophy behind 'multiple pass compilers'. It greatly helped me read many papers and stay informed about new ideas in the field (although I can only reproduce formulas and proofs from a handful of them that I found relevant to my research).
Short point summary
- Read a lot of papers
- It may be helpful to keep track of which papers you read and what were their key claims
- Try 'multiple pass reading' technique