Showing proficiency in a subject without taking a course

Question

I want to communicate to graduate admissions committees that I know the equivalent of a standard course in machine learning as a result of my self-study. If possible, I would like to avoid taking the machine learning course at my home university to prove this.

Is there any way I can show my proficiency in the subject matter? There are no proficiency tests in my college for this course.

I have a few ideas:

1. I could blog about machine learning
2. I could post solutions to exercises from a machine learning textbook
3. I could make a project that uses machine learning
4. I could make YouTube video lectures/tutorials.

How would graduate admissions committees compare the following with a good grade in a machine learning course?

---

Rationale for not wanting to take the course (Please do disagree if you see necessary).

I think self-studying is more efficient than taking a course. There's a lot of overhead involved in going to class, submitting assignments as per a certain format, clarifying grading policy details etc. that I would rather divert to learning more of the material.

That said, having read one of the links in the comments, I now see how unverifiable self studying can be; many students can allege to have proficiency but be dead wrong.

---

Answer 1

If showing proficiency in this one area is so essential to your application for graduate study that without it you would fail, then you are already in trouble. It would be different, of course, for a required undergraduate course such as Algorithms. If your general record is strong without this, then don't worry too much about showing proficiency for application to a program. You can mention in your CV or elsewhere that you have also studied machine learning on your own and it will be noted as a plus, even if unverified.

The problem in general with evaluating self study is that it can vary so widely. It is hard to verify that you've studied in sufficient depth, but also that you have covered the ground that would be expected in an UG course. Of course, if you have written machine learning algorithms and tested them, then your code and results can speak for the depth, but not necessarily the breadth.

However, if you want to study machine learning as a dissertation topic, then you need to convince one person, your advisor, that you have the necessary background. You can probably do that in an interview, which might have some aspects of an informal oral exam.

In general, though, some advice to self studiers. Don't think that you have learned something just because you have read about it or were a relatively (or completely) passive participant in an online course. Solve a lot of problems, and save your working papers. Build yourself a portfolio, which can just be a stack of paper, that shows that you have engaged successfully with the subject. Attend to the breadth issue. If standard textbooks exist they can be a good guide to breadth and their exercises and suggested projects a good verification of depth, provided that you do them.

Answer 2

In my opinion (and the answers to this question will be opinion based!), 1, 2 and 4 are not helpful: 1.: Everyone can blog about a topic they have no idea of. And no committee member wants to read the full blog. 2.: Noone wants to verify whether you did it right. And no one can be sure, that you did them by yourself. 4.: See 1

3 has the problem that a) it might very likely just cover a certain subset of the whole material and b) no one knows, if you did them by yourself.

The only way in my opinion would be a research paper with you as main author accepted in a peer reviewed conference or journal.

I would just try to be honest and write which courses you did and how you applied the material to practical problems - without trying to prove any equivalent to formal courses.

Answer 3

I doubt a graduate school cares about some specific individual course. Of course if your actions mean you won't complete the B.S. that is another matter.

I wouldn't do all the elaborate stuff (videos and the like). It will look silly. Just a single sentence (within overall essay and without disproportionate emphasis) saying that you did an organized self study of topic X and thus have the benefit. And no need to go on some tangent explaining why you did this--realize the people taking your application are professional, paid instructors. Personally I agree with your point of view. But it's just not the right place (applications essay) to be making a contrarian education argument. Nobody cares, it's not germane, and it will turn people off.

[But then you SHOULD have an organized self study...good text, sound projects, and some means of assessing yourself..even with an alternate text's problems or with a tutor (or "last year's final" or the like).]

Answer 4

Consider taking the course.

I don't mean to devalue your knowledge or assume that you did your studies half-heartedly. Most of my skills are self-taught with the help of online courses, so I reside firmly in the "for some people, self-study works better" camp. But based on my experience, taking a course that seems to cover something you have already learnt, will still offer some benefits:

1. You revisit knowledge, repeat difficult elements and thus improve your overall proficiency of the subject
2. No two courses are alike. It is quite likely that it will include things that you haven't learnt, or that you have only superficial understanding of. Especially since you are self-taught, you might not have been confronted with issues that arise in a classroom context: collaboration, tools, adherence to standards, ...
3. Use the opportunity to help other students and learn by teaching, again consolidating your own skills

Standing out: Approach the lecturer and ask for more in-depth detail of the course. Explain your situation. Instead of trying to escape the course, offer your support (if you really exceed the course requirements by margins) or ask for additional material (if you are certain of some things, but not all course requirements). I teach an introductionary programming course to students, and the studentship is extremely heterogeneous: from "I googled programming and decided to try it" to "I am currently working on my own game in Unity". As attendance is compulsory(!), we try and encourage the advanced students to complete extra challenging material, to help other students, or to spend the classroom time otherwise, but still related to the subject (i.e. no gaming or browsing). So you would still take the course, but you would have quite a bit of leeway in spending your time more effectively.

At the end of the day, you have to ask yourself: what do you gain by skipping these x amount of hours, what do you lose (also consider the social aspect: students grow together, yet you're away) and instead of fighting the system of admissions, you could view it not as a penalty for prior knowledge, but as an opportunity. Maybe you will meet others with prior knowledge (not unlikely in this subject area) and perhaps you are less knowledgeable than they are.

I say all this because I have been there. I am largely self-taught, and have found myself in situations where my knowledge was not more or less, but skewed. While I had experience in implementing some very advanced features of some programming language, I was, at the same time, missing important basics that I had never come across/hadn't needed. It not bad, it is just that there is a mismatch. Taking the course for you is an opportunity to gain knowledge. You won't lose any.