Can an undergraduate publish a research paper?

This post is specific to most of the engineering undergraduates which are interested in research but do not know how to direct their efforts. Since my specialisation is in Machine Learning I would mostly cite examples from ML, but don't worry they are pretty standard and easy to understand examples.

First of all, a little about me: I am a final year engineering undergraduate from a tier 3 college in India. I have authored one paper in Springer affiliated journal and 2 papers in flagship IEEE conferences. I will add the links to my papers once they are online(:P)

Motivation: Many undergraduate students want to publish a research paper in order to explore the field of their interest or to improve their profile for Masters admission. Whatever might be the reason, one of the main problem every undergraduate student from a tier 3 college faces, is the dearth of research assistantships in colleges. Many colleges do not have a great research culture and as a result, a student interested in academia can suffer. My main aim is to share my experience and help people with their research journey.

That’s a lot of talk. Phew!! Now let's get to the point.

Step 1: Pick a field you like

Now, this can be any field such as VLSI, Machine Learning, Image processing anything.

Pick a field of your choice and if you don’t know which field you like, pick the one you find the most intriguing.

Step 2: Choosing a sub-specialisation

Now that you have picked a field, chances are that the field is very broad and there are many small specialisations in that field.

For example, the field of Machine Learning is vast and versatile in nature. It can be used with image/signal processing, for making predictions, for estimations and for generating datasets!!!

The point I am trying to make here is that it is important to not get overwhelmed. At this stage, I would suggest finding a faculty.

Every faculty member is working on their sub-specialised domain. Pick a faculty that works in a similar domain as the one you selected in Step 1 and approach them. For example, if you are interested in Machine Learning you can just go ahead and approach the professor that taught you Machine Learning.

Once you had a conversation with a faculty, they would be happy to help you. They might share with you their problem statement or least of all they will tell you different sub-specialisations that are available in your chosen domain. I had chosen machine learning as my domain and choose signal and image processing as a sub-domain in the field.

Now that you are aware of the sub-specialisations in your field, try to gain some basic knowledge about your favourite sub-specialisation. Google is your friend at this stage. Try to do multiple online courses and get a hang of your chosen sub-specialisation.

You can do a lot of trial and error at this stage. experiment a lot. Try to learn multiple different things in your chosen domain.

Step 3: Intensive Literature Survey

This is probably the most important step in the process.

After following the previous steps you must be clear about

Once you are comfortable with the above-mentioned points it is now time to find research papers. Since you are clear about the sub-domain it would be easier for you to find the papers. For example, if my domain is Machine Learning and my subdomain is signal processing I would search “Machine Learning applications in Signal Processing” on Google Scholar. This would render thousands of recent papers relevant to my sub-domain

Collect around 15–20 recent papers(not more than 6 years old) and store them in a folder sequentially in ascending order of date of publication.

After making such a list, skim through all of the papers and just get a basic gist of the papers. This should not take much time. There are two advantages of this step.

Step 4: Get your hands dirty

By implementation, I mean actually recreating the results mentioned in the paper!!! You have to write the code and implement the entire paper from scratch. You can ask the authors of the papers for help if you are stuck. Recreating the results is probably the toughest stage for a budding researcher.

The main thing at this stage is to not give up. Keep on trying!!!!. Keep on struggling!!!! Remember no one can learn swimming by watching Youtube videos. Get your hands dirty and don’t give up until you successfully replicate the results of the chosen paper. It may take months (took me 7 months), but the fruits of that struggle are immeasurable. Perseverance is the key here.

Step 5: Find your problem statement

Congratulations !!! you have done all the hard work.

After implementing the paper you would have understood why the author proposed the current method and the impact of that method. Now we have to do just one thing, that is to find a loophole in the paper we just implemented.

It might sound challenging on the first go but trust me it really isn’t. Since we have implemented the method proposed by authors from scratch it is going to help us in finding a loophole in the suggested method.

At this stage, it is important to realise every method has certain flaws in it. Every research paper has one small caveat where the proposed method would not work. It is important to find this loophole.

Once you begin this trial and error phase trust me you will soon find a small caveat in the paper. Again perseverance is the key!!!!. There is no perfect method and you can find the loopholes. You just need good googling skills and a lot of perseverance.

Step 6: Finding a Solution

This is the easiest of all the steps. Once you have found a problem in the implemented paper finding a solution is a piece of cake. This is because, due to extensive googling you are already aware of the state of art methods available in the market and the pros and cons of each and every method.

Your knowledge base becomes so vast till the time you reach this stage that proposing a solution is a piece of cake. It is just about putting your thoughts into code and checking if your proposed method is solving the proposed problem effectively.

Just remember that your solution does not need to be completely unique. It is completely fine if someone has already suggested your solution. Just make sure that the suggested solution has never been applied to the problem you have just found out!!! For example, I had suggested a CNN named classifier in my paper as a counter to the problem I had found in the previous papers. Now, I did not invent CNN. I just suggested that CNN could be used to tackle my problem.

Step 7: Documentation

This step involves documenting and formatting your proposed methodology. I will not give you many details on this as there are many tutorials available online which can guide you through this step. Just make sure that the paper you write is lucid and easy to understand.

I have given a 6 step guide on how an undergraduate can publish successfully. If you have any queries you can contact me through

Remember one thing that it is a process — long and arduous one for that matter. It took me 1.5 years to get my first paper. Just make sure that you don’t give up. Continue to learn, continue to struggle!!! All the best!!!

Related posts:

Kenan Kodro er ikke typen der kom flyvende ind på scenen og brændte den af fra start af, men skulle bruge nogle år på at løbe tingene i gang. Først mod slutningen af 15/16 sæsonen i Osasuna, så han…

My name is Father Dmitri. I have been an accountant for the Catholic Church in New York City for the past fifteen years. On that recent fateful morning I had run into a boy named Angel Martinez…

Most people think ignorance is a bad thing. But ignorance can be a good thing. Ignorance can give you an edge. I’m not talking about being ignorant of basic science, or ignorant of other cultures…