2
$\begingroup$

I wanted to ask this on the meta before going all out on the actual SIGPROC site. What I would like to know is how is one supposed to go about asking Sigproc questions that related more to breadth than depth.

The context here is that I would like to ask questions on a topic, but am not really aiming for such an in depth dive into it, as much as a 'breadth' insight on it, what it is used for, limitations, what people have experienced with it, etc. Now, each one of those can be or become a question onto itself, but it would seem that posing a dozen related questions on the same parent topic can get out of hand and tedious, plus context will be lost in that regard. So how to go about it?

As an example, (I actually want to ask about this but here is a preview), I want to ask about how "Maximum Likelihood Estimation" is used in the context of signal-processing. So I would like to know (briefly) what it is actually doing, here the math and detail would be important of course, perhaps some information on how to actually implement it. But then from there, I would like to know when do us signal-processing tend to use it? What are its limitations generally speaking? When is it good to use it? When is it bad to use it? What are cursory topics related to it that improve it that one might be able to look into?

So you see, I am looking for a 'stake in the ground' on the topic itself, but then some 'fan out' around it to get an idea of the lay of the land, and I am wondering what if any special formatting I need to do to make sure it is kosher for the sig-proc stack exchange. Shall I put all that into one question which will have many sub questions, or divide all of them up into their own distinct questions?

Thanks for your inputs!

$\endgroup$
6
$\begingroup$

I may be wrong, and I hope more people will respond to this, but I don't think that a question titled How is Maximum Likelihood Estimation used in signal processing? is precisely appropriate here. It is something one can easily give to a search engine and get days of fantastically interesting reading as the result.

Stack Exchange web sites, at least the technical ones like Signal Processing, exist for problem solving and research assistance, but not primarily as educational resources in the broadest sense. So if the question were of the form Can someone help me understand the reasoning behind using maximum likelihood for estimation? and came with a few examples of what exactly the OP is confused about, people will gladly jump on it and answer the question.

There are two issues to tackle here. First of all, there is no single correct answer to your question. The most correct response would be the one containing the most references and topics. Therefore, your best chance for keeping the question on the site after moderator review is if it gets converted into a community wiki (CW). This way people will simply augment each other's responses and improve them over time.

On the other hand, people with expert knowledge mainly value their time as quite precious, and many of them will be insulted, for the lack of a better word, with how little effort the OP put into asking the question.

I'm really trying not to make this response negative, because you're bringing up a very, very valid question, and I think how we answer it matters to how we will react to very general questions in the future.

Update:

To respond to the issue of breaking up a question into several smaller ones, that should be by all means fine if the following two conditions are met:

  1. The OP has put effort into researching the topic first, and questions are specific to certain understanding the theory or implementation of a particular idea or algorithm. In other words you're not simply taking a large topic and trying to make people teach you the topic step by step.
  2. The questions are independent enough so that if one question is answered, you would still look for all other questions to be answered as well.

In any case, I don't envision many scenarios where one would have to ask several questions on a certain topic at once.

$\endgroup$
  • $\begingroup$ Thanks and I see what you are saying and yes its certainly something important as you said. Dont put too much weight on that particular question btw, it was just an example, but you can see my drift... :-) Anyway, I can break/phrase such a question into a technical one... I have technical (as in relevant to our SE) questions related to the topic (ML)... that are sort of related but can be questions in and of themselves... if we were all sitting around a table this wouldnt be an issue but wanted to follow whatever SE rules/expectations there are. :-) $\endgroup$ – Spacey Nov 22 '11 at 21:42
  • $\begingroup$ (cont) The relationship between the various sub-topics of a main topic in this case would be technical for example, (I would make sure it is at least), but I am wondering how/what formatting would be needed. Perhaps I can try and post it anyway and we can see how it goes. Anyway thanks for your input for sure! $\endgroup$ – Spacey Nov 22 '11 at 21:44
  • $\begingroup$ @Mohammad Yes, that certainly makes sense. I also updated my response. $\endgroup$ – Phonon Nov 22 '11 at 21:52
  • $\begingroup$ Awesome thanks Phonon! $\endgroup$ – Spacey Nov 22 '11 at 22:29
  • $\begingroup$ When learning about new techniques some of the 'fantastic reading material' may be a bit daunting and leave you with valid questions. So the problem is: how do we help people ask the right kind of question, so they get an answer and we get a great question. $\endgroup$ – Ivo Flipse Nov 24 '11 at 11:29

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .