Discussing sole author vs first authorship in ML/CV with supervisor

Question

I am an UG research assistant in CV/ML on a robotics project. I was given a surveillance video by my boss, and w/o being too specific, I was asked to find a way to detect something in the video. W/o revealing myself by giving too much info, I went above and beyond and found a way to predict something occurring in the videos. I did everything from collecting the data to implementing the code. I am not sure what I should do.

Usually, a co authorship or first authorship assumes some kind of mentoring in UG. There was 0 mentoring or guidance, so this would be wrong. I was given a project and left alone to do it. At the same time, he took me on and gave me a chance with little experience. Additionally, I would want to use him as a LOR, so I don't want to upset him.

Short summary: I discovered a new and effective way to do something that AFAIK, has not been done before. I am not saying this is groundbreaking research, but it's not too bad for UG work. I spent a little over a year on this project and worked extremely hard. I received no help. I am not sure if I should ask my boss if I can solo author. Should I write a draft of the paper, then ask him if I can solo author? He supplied me with a video, in which I collected data, can I acknowledge him in it instead of coauthoring?

Answer 1

Just go with co-authorship

You're an undergraduate researcher. This is a case where having co-authorship is actually better than sole authorship. You said yourself it isn't groundbreaking. By doing co-authorship you've created a google trail back to this professor.

When people are googling the professor's name, your paper will come up. When people google your name, they will see your research network back to the professor.

Ph.D. students in the group likely talk about first/sole authorship more. That's because Ph.D. students are expected to do high-level research and come up with novel ideas. The next step in their career is (maybe) being a professor.

The next step in your career is getting into grad-school

The prestige of the professor your working with is a factor in admissions. He'll probably write a recommendation anyway, but drive it home to admissions committees with a co-authorship.

From your post it sounds like he's pretty hands-off. Write up the paper and put your name first and the profs second. Let him tell you to swap them.

Answer 2

The co-authorship route is probably better, and there are several benefits you would get from him. A scientific research article is written in a different way to most other types of articles, so you would be getting mentorship in scientific writing. Without this mentorship you may have much more difficulty in getting the paper accepted due to your unusual writing style, the fact you will likely place emphasis on the wrong things and leave that you may leave out key points that the readers need to follow.

Your supervisor also knows the field better than you, and so knows if this is a publishable result, or if you will need to do some checks/extensions. They may also realise that your method may be an extention of some other method or an application of a method to a different field (even if you didn't know of the method before). That may give some insight to the result and/or make it fit more snugly into existing literature. It may also turn out that your result was published last year, or ages ago in some out of the way place, which your supervisor may be more aware of.

There are also money/grant aspects to consider. Money may be involved in getting it published (though it might also be possible for the university to have some money, but they probably expect post-docs/advanced PhDs to be requesting it). I have also heard that computer science papers are normally published by going to a conference which will have expenses. If you got some money from a grant your supervisor may wish to say that this is an output of that grant so they can include it in reports and it may look weird not to have their name on it (shouldn't really be reason for name on a publication, but might be useful if trying to understand any behaviours).

All in all: probably go co-authorship, but view it as a mentorship in how to write academically and instruction in how this work fits into the broader picture. However you can discuss with him about being a corresponding author (which may indicate you are more than just an undergrad doing plug and chug who got lucky). Furthermore when you get a reference from him you can ask him to explicitly point out how you took initiative and went beyond what was expected, resulting in this paper.

Answer 3

Contrary to some of the other answers, Id suggest that you have a conversation with your advisor about it. Don't make assumptions. But ask what the advisor thinks what would be appropriate here, putting them on notice that you have thought about the issue and have some preferences. But making any assumption without having this discussion could lead to problems and you might wind up with less than you deserve or hope for.

Ask whether you should be sole author with an acknowledgement for the professor's help in getting you started, or whether co-authorship would be more appropriate and helpful to you in the future. If they make a recommendation, ask for the reasoning behind it so that you can learn more about how academia works. Listen to their answer and analyze what it means.

Don't bring up the issue of stealing your work. If the professor is someone who would do that, then they might anyway, but, while it happens, it isn't as common as you fear. Yes, there are a lot of questions here from students it has happened to, but far more students have no need to ask.

I don't think you will lose anything in starting this conversation. The professor won't suddenly think "I can steal this work." Trust, but verify.

But, as the other answers suggest, co-authorship can be a good thing. Don't reject it out of hand if it is suggested.