21

As a student starting his masters degree, I was given an academic textbook on reinforcement learning, and I was asked to read it by next week by the professor. I just joined his research group.

I looked at the book and it seemed obvious to me that I couldn't hit that target, even if I could spend the entire day on it, for all 7 days. I know a bit about this field, but not enought to finish this book in 7 days.

Is that an indicator that I am not cut out to work in his group?

11
  • 25
    You can read textbooks at very different levels of detail. Do you know what level is expected from you?
    – Anyon
    Commented Jan 16 at 21:31
  • 86
    @sage76 Lesson (hopefully) learned: next time you are given a seemingly impossible task, ask your professor/advisor what it is they actually want you to do instead of trying to divine their intentions with the help of internet strangers. Commented Jan 16 at 23:55
  • 40
    @sage76: You've complained about a couple of different comments now. While I see how Adam's comment might seem condescending, I think he was really trying to help by telling you that, in future, it's totally appropriate (and in fact, better) to ask questions like this one to your advisor. If there's some reason why you don't think that's a good idea (e.g., because your advisor will scream at you), explaining your reasoning here could help people give more accurate suggestions.
    – cag51
    Commented Jan 17 at 1:23
  • 14
    FWIW, Adam's response read as unnecessarily condescending to me. And the OP's reaction a bit aggressive ¯\_(ツ)_/¯ Commented Jan 17 at 8:52
  • 26
    I did not see @sage76's response to my comment, but gee, what a strange coincidence that every time someone writes a fake RateMyProfessor review for me or signs my email address up for a million spam lists out of the blue, it just always happens on occur on the same day that someone anonymous user gets very offended by one of my Academia.SE comments (which was meant to help you, sage76, deal with this type of situation next time it arises). Thanks for commenting, guys, otherwise I wouldn't figured out what's going on now that the OP's comments have been deleted. Commented Jan 17 at 13:58

3 Answers 3

49

I'm not in your field, but I think I have an understanding of the sense here:

It's absolutely NOT "read this the way you'd read a textbook for a 9-month course which would have quizzes and tests..."

It's "here's a body of info, that you need to be acquainted with... and can look back at references AS NEEDED".

The "read this book over the weekend" is an exercise I occasionally give my PhD students in math (US, at R1), not only to get them to look at the stuff, but to prove, conclusively, that any ultra-pedestrian, pedantic, approach to getting up to speed is doomed. Thus, a different concept is necessary. :)

(That is, seriously, when doing what one thinks is necessary is impossible, it's time to reconsider...)

8
  • 1
    I'm a bit curious about this in maths. Im finishing a math masters but currently Im dabbling (and heading) towards a very interdiscplinary field. I can get acquantained very quickly in the interdisciplinary field, because really the math present is quite trivial after 5 years of studying. I however have never done serious research in mathematics, and how does one "skim" a mathematics text? AFAIK, math is discerning the defined objects and understanding the proofs, how can that be skimmed only? Commented Jan 17 at 22:05
  • 2
    It's true, most of us have been trained to think of the essence of mathematics as being lots of delicate details... but, I'd claim, that stuff is really just an artifact of a style of exposition. Mathematics is about phenomena. :) Commented Jan 17 at 22:21
  • 12
    @paulgarrett Definitely agree. The pedantic details are necessary---it isn't true if you can't prove it---but there are only about half a dozen papers in mathematics that I have really read in detail (and each one took somewhere between a month and a year to really grok). But I have read hundreds of papers to the level of "Oh, I see what y'all are doing. That's pretty nifty." That only takes a couple of hours. Commented Jan 18 at 1:51
  • 4
    Giving people (and yes, students count as people) work over the weekend should be illegal and looked down upon, and not glorified.
    – Davor
    Commented Jan 18 at 14:21
  • 5
    @Davor, the only people I do this with are my research students with whom I have a long-term (good) relationship. And their research is for their own benefit, not mine. So they always can decline. I don't "give orders", I give advice, based on experience. Sorry if there was a misunderstanding. Commented Jan 18 at 18:14
28

Writing this answer from the perspective of a physics graduate student, which may be different from CS.

You have joined a research group, and you have an advisor. This is a fundamentally different relationship than a boss in a job. A boss will say "do this" and you must do it. With an advisor, they say "do this" and then you report back during the next meeting whether you could do it or not. Modestly reporting that you couldn't do something is normal; truly new research has an unknown outcome, so failure is often the expected outcome.

That is, the correct course of action here is to report to your advisor that you will need to take more time to understand the material in the text. From this, both of you will have a better idea of your ability to tackle different projects.

Here is my advice:

  1. Skim through the book, and make notes of what you do understand and what you do not. Make special note of things that you find interesting.
  2. Estimate the time to master each topic
  3. Put it into a powerpoint presentation, with a length of about 5 minutes.
  4. Present the powerpoint at the next group meeting.

After this conversation, ask for feedback. You advisor can either tell you what specifically to focus on in that book, or suggest a different book. Or work with you to find a research project. Or something else.

5
  • OP should accept this answer, as it is the most professional course of actions to be taken. Extremely helpful advice.
    – The Doctor
    Commented Jan 17 at 15:55
  • 13
    A PowerPoint seems like overkill to me, but then I never liked PowerPoint anyway— I would just say something along the lines of "I wasn't able to read the whole book thoroughly yet, but I reviewed the high-level concepts and have a good understanding through chapter X. I did have some confusion around concept Y, maybe we could discuss that later? I could probably finish a thorough read-through by Z. Are there any parts that I should prioritize or give extra consideration to? &c." Obviously, omitting parts that are inapplicable. Commented Jan 17 at 21:03
  • 5
    If you don't tell your "boss in a job" when a demand is either impossible or unclear, then he will yell at you next week when you return with half a result or haven't finished other tasks because you overdid that first one.
    – Karl
    Commented Jan 17 at 23:23
  • @A.R. Yes, but a powerpoint it is a great way for a new student to talk about their attempt to accomplish a difficult task. Commented Jan 18 at 1:34
  • 3
    If anything, from my experience, academics are far more entitled then any "boss" I've had in the industry. I've never seen the level of idiotic deadlines given to me in education compared to now working in tech.
    – Davor
    Commented Jan 18 at 14:23
8

Just to expand somewhat on @Anyon's comment:

"You can read textbooks at very different levels of detail. Do you know what level is expected from you?"

Taken literally, "I was given an academic textbook on reinforcement learning, and I was asked to read it by next week by the professor.", your professor has only asked you to read the book. Just reading doesn't take long. If you read 50 pages each morning and each afternoon, then it ought to be easy to read a 400 page textbook in a week (and if it is the excellent book by Sutton and Barto, then I have read it in about a fortnight of reading in my lunch hour, which is a comparable amount of time).

Just reading a textbook is a valuable thing to do - it ought to be sufficient to give you a good understanding of the key concepts, terminology, conventions, problems and ideas. That will give you a common frame of reference for discussing the topic in greater detail with your professor in a very efficient manner. So this doesn't seem an unreasonable request, especially if it is your only assigned activity for the week.

I suspect the mismatch in expectations may be that you think the professor expects you to have fully digested the material and understood the mathematics in detail, but I doubt that is the case - it would definitely take more than a week!

A few years ago, I was assigned to teach the operating systems part of a computer architectures and OS course (which I hadn't done since I was an undergraduate back in the 1980s). So I read Tannenbaum and Bos (1136 pages), Stevens and Rago (1013 pages) and Kerrisk (1552 pages) cover-to-cover as well as sampling some other good OS course texts. The reason I did so was not to gain an understanding of everything covered in those texts, but to give a broad overview of what there was that might be worth covering in the course and what the API provides for programming activities that I might use as examples, demonstrations and coursework. It also meant if a student asked me some obscure question, I would at least have a vague idea of what they were talking about and roughly where to find the answer. I got that just from a month or so of lunchtime reading. Time well spent - it enabled good communication with my students.

4
  • @downvoter an indication of where you disagree with my answer would be appreciated. AFAICS I have just suggested a plausible explanation of events where the request seems reasonable (and useful for student-supervisor discussions). Commented Jan 19 at 16:21
  • I am not sure if you got what "reading" meant. If I was asked to read a book, I would read the main text body, but skip proofs and technical details. Large parts of the book would not actually be read.
    – usr1234567
    Commented Jan 19 at 19:21
  • 1
    @usr1234567 that would be one level, I was more thinking of looking at the equations and technical details as well but not taking the time to fully understand the derivations, just the principal results (to make sure they agree with the intuitive understanding of the topic gained from the text). They key point was to learn enough to engage in a discussion on the topic (which may require familiarity with the mathematical representations), rather than be an expert. I need to read a book on tensors in stats, I will start with that sort of read before going into details. Commented Jan 19 at 19:26
  • ... because it will mean that my intuition will be better when trying to understand the details of the maths. For me the key to maths is intuitive understanding (usually the physics), whereas Mrs Marsupial approaches maths in a much more symbolic way. Different people will have different methods, but the equations is where we both find the common frame of reference, so I do need to be familiar with that as well. Commented Jan 19 at 19:33

You must log in to answer this question.