Background: Neuroscience, India

My thesis is based on the use of machine learning for some neuroscience data. Since I do not expect the readers to be aware of machine learning details, I have written a high level summary (a few pages) before the literature review which summarizes the basic principles of machine learning (like introducing what is cross-validation, how to measure accuracy, false positives, etc.). The idea was to let the reader gain some high level understanding before diving into the actual literature review so that the reader would be able to better appreciate the work previously done and the knowledge gap.

This part of the thesis is based on my understanding, clubbed with courses on machine learning and generally reading around on the internet. As far as I can see, the ideas are simplified 101 which can be easily found from Wikipedia (such as the page describing confusion matrix) and from online blogs. To summarize, as far as I can see this is borderline common knowledge.

My thesis reviewer has raised a concern with this part saying that I have not cited any sources and I should cite academic sources for these concepts. I don't quite see how that is possible as the concepts I have described are very elementary. Most textbooks that describe these ideas do not cite any papers and I don't even know how I am supposed to find citations for concepts like precision and recall!

Question: is my understanding of "common knowledge" incorrect? How could I reply to this concern/find sources for these introductory concepts?

3 Answers 3


Going to frame challenge a bit here...the situation you describe in your question doesn't really fit together with the problem you're experiencing.

Either you refer to things "common knowledge" - in that case, really not much need to cite but also no reason to put in your paper as its own section, or you refer to things "not common knowledge" and you provide citations.

You'll have to decide whether this review of basic ML concepts is overly pedantic and underestimating of your audience (I tend to think it is based on your description; you should not be defining 'confusion matrix' in a thesis), or if it's advanced enough to require citations throughout.

One hybrid approach, and the one I would take in your situation, is to free yourself of the super basic ideas like "cross-validation exists and here is its definition", and instead discuss for your particular implementation the nuances of which variety of cross-validation you are using and why it is important to do so. There are certainly review/primer-style articles in academic journals published recently on this sort of topic, as authors argue for one approach or another.


You are being a bit inconsistent here. The reason for the introduction is that readers may not know the background. It would be good for them to see some sources that will fill them in beyond what you say. Mentioning a standard text or similar might be enough.

You also have risk with the reviewer if you don't yield to them. They may have a lot of control over what they accept. Again, pointing to some standard material, would probably suffice. If not, they will hopefully tell you to say more.

While I think you are probably correct about your interpretation common knowledge and the lack of requirements to cite it, the other factors suggest you should do so in any case.


Cite repeatedly from the Encyclopedia of Machine Learning, available here: https://link.springer.com/referencework/10.1007/978-0-387-30164-8

It's a reference work, where many famous people in the field explained core ML concepts in an accessible way. Cite whatever lemma you need for whatever concept you introduce. The confusion matrix is definitely in there.

This way, you satisfy your reviewer, you avoid charges of not embedding your work in the relevant literature, but you also clearly show that the concepts are of a fundamental level by referring to an encyclopedia.

  • Thank you for this reference! Very valuable!
    – stuckstat
    Nov 21, 2020 at 14:32

You must log in to answer this question.

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