31

I am a student doing my Masters (in Computer Science). I am much interested in Ph.D and I know to do Ph.D I really need to read and understand a lot of papers. (I also know that its not all about reading papers, but doing something new and useful) but I find difficult to understand papers, especially mathematical. There are lots of things (like bloom filters, k-means algorithms, and a lot lot much more terminologies) which we, normally, don't encounter (except if you are a Ph.D or doing one) but are really crucial to understand papers.

I am not very good at understanding such stuff (atleast quickly, if given time I can but it takes time). Are there people like me who have done/are doing a Ph.D and had the same problem? If yes, how do you overcome it?

  • There are great books around which will bring you up to speed on a lot of general techniques e.g Pattern Recognition and Machine Learning by Bishop, Machine Learning: a Probabilistic Perspective by Murphy. There are great written tutorials online made by people to explain specific techniques such as k-means). Also as you are in CS papers can be accompanied by code which often assists in deciphering technical sections of the paper. – Stephen Tierney May 15 '13 at 3:03
  • 1
    It took me a month to read and understand the first paper I needed in my research. Now, 5 years later, I can skim through 4-5 papers in a day, and follow the ideas. The key is practice, practice, practice, and read papers that are less technical first, that gives the background. – Per Alexandersson Mar 16 '14 at 18:48
35

The good news is that you have a good 4-6 years of time to learn how to process that sort of material, and you aren't expected to be able to do it immediately. A professor once told me (about physics papers, but it translates into mathematical and CS papers fairly well) that eventually you start to read the math-heavy parts a bit like music -- you see familiar patterns, and you get the general idea without having to dig deep into the details.

Additionally, you'll find that the more classes you take, and the more reading you do really does prepare you for that next paper and that next step. You'll find that as you start specializing on a particular topic, most of the papers in your topic will become easier to read, simply because you have the experience and have gained the knowledge about the particulars over time. Furthermore, you'll start coming up with your own ideas in the subject, and this just happens to be the goal of graduate school!

There are lots of things (like bloom filters, k-means algorithms, and a lot lot much more terminologies) which we, normally, don't encounter...but are really crucial to understand papers.

The beauty is that you will start encountering them more, and you'll start understanding them more and more as you continue your education.

To answer your particular question,

Are there people like me who have done/are doing a Ph.D and had the same problem?

I still have difficulty getting through math-heavy papers (in CS), but I either re-read sections closely until I figure it out, or I ask someone to help, or I simply move on and hope that the rest of the paper fills in the details. I don't expect to understand everything technical that I read, but it becomes easier over time to get what you need out of a particular paper even if you don't get each and every detail.

  • 10
    +1 for "eventually you start to read the math-heavy parts a bit like music" :) – d.putto May 10 '13 at 10:17
  • 1
    Agreed, what you get better at really is figuring out what you should spend the time to truly understand. – KennyPeanuts May 10 '13 at 19:48
16

I don't think very many people come into academia with the ability to parse tons of math-heavy articles quickly. It's a skill that develops over time from reading lots of papers, asking someone in the know to help you understand confusing parts, and getting through relevant coursework that will build up your skills and intuition. If it's a matter of speed of getting through things, that will undoubtedly improve in time. If it's a matter of you simply not understanding them at all, then you may want to reconsider your options before diving into a PhD.

8

Start by not reading papers straight through. Most research papers are difficult to follow if you read through start to finish without a loooot of context.

Normally, I read them in approximately this order to begin with:

  • Abstract/introduction
  • Conclusion/summary
  • Future Work
  • Results
  • Methods

You want to understand what the paper is about first. Generally speaking, unless I know a topic well, the beginning section doesn't do a good job answering the, "what is the point of the next X pages" but a conclusion and summary do. Future work helps me understand, "why is this research even relevant" and then going backwards results helps me understand what they did, and then, once I have context, I read through methods.

Honestly, too, I don't care about understanding detailed methods unless I see a value in understanding them. Reading for the sake of reading is hard, so I "incentivize" myself by showing "oh, there is value in understanding the details of what they did!" and basically trick myself by how I read through things.

Depending on how well I must understand a paper sometimes I stop early in that process or reread sections, etc.


In addition to this, make sure you don't read every paper word for word. You don't have to.

Also determine if you are able to read on your computer or if you need to print them off. Some people cannot effectively read on a computer. Some can. Find out your personal abilities here.

  • 7
    OP specifically mentions mathematical papers. Mathematics papers almost never have "methods" sections, and rarely have separate "conclusion" or "results" sections (because the entire paper is results, and the results are the conclusion). – JeffE May 10 '13 at 13:54
5

If it helps, I was two years' into a full time research degree in psychology (no courses) in which I read all day long and one day I suddenly realised that I was reading an economics book with very little effort. I could read! I still feel the excitement. You do just get better and better but it is like going to the gym - constant application is needed.

Writing took much longer to get easier and I strongly recommend that you make a habit of writing daily. Make sure you always have a writing project on the go and add something every day, no matter how bad. Get a calendar; cross off the days; and don't break the chain. You needn't spend long - under an hour will do. But constant application is the key.

If you are lucky, you will have mentors who demand papers from you. It is a lot easier to write if you have deadlines. But if not, start a blog and start building up a portfolio. The feeling of pressing the publishing button and seeing your blog grow makes up for disinterested supervisors.

Stick with it - in few years you will marvel at the ease with which you find your way around 'heavy' material.

1

I have found that reading papers of a specific topic within my research, then looking at Google Scholar to track down similar articles helps. I always have a relevant dictionary handy and write down questions related to the topic that I am not to sure about for follow up.

Taking regular breaks to reflect on the information is also useful.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.