Arrykrishna Mootoovaloo

How I Take Research to Publication

The steps I follow to take a research idea to a first-author paper.

2 min read Machine Learning and Statistics

In this post, I outline the steps I have found most useful in carrying out research and taking it through to publication. The first first-author paper is generally the most challenging. It was through this process that I learned to think critically and to answer questions such as 'Why is this research important?', 'Where will it be used?', 'How does it differ from previous work?', 'What are its new contributions to the field?' and 'How do I manage my time effectively?'

  1. Project definition

    2. The first step is to define the project. There is, however, always an element of risk at this stage: researchers sometimes find that their projects no longer seemed promising after several months of work.

  2. Brainstorming

    3. To avoid the pitfalls described above, I strongly recommend holding brainstorming sessions with your collaborators to identify possible risks and limitations.

    "If I had an hour to solve a problem I’d spend 55 minutes thinking about the problem and five minutes thinking about solutions." - Albert Einstein

  3. Identify a main reference paper

    4. Finding a main reference paper is fundamental to the whole process. This means understanding the paper and its mathematics, and being able to implement (code) at least part of it.

  4. Fill in the gaps

    5. At this point, we typically have only a blurred picture of how to develop the main idea, largely because of gaps in our knowledge of specific topics. It is very helpful to master these topics (through textbooks, lecture notes, online resources, YouTube and GitHub) to fill in the gaps.

  5. Experiments

    6. This is probably the most important part of the entire process, where we write and test our code. It can seem daunting, as we will inevitably encounter many failures. However,

    "Failure provides the opportunity to begin again, more intelligently." - Henry Ford

  6. Updates

    7. Crucially, it is good practice to document our code and draft the paper as we go. This allows us to keep track of everything we have done since the start of the project.

The Paper

Different journals require different formats, so it is important to identify and adopt the right format from the very beginning to work efficiently. Below are some brief suggestions I received from my supervisors:

  1. Abstract: Summarise key findings. Numbers are essential.
  2. Introduction: Motivate the study.
  3. Body: Elaborate on the theory, data, methods and results.
  4. Conclusion: Remind the reader what the paper is about and briefly explain how its aims have been met.

Most Important Suggestions

  1. Keep in regular contact with your supervisors (emails and weekly meetings).
  2. Have at least one mentor. I have found it easy to talk to my collaborator, who is a Research Fellow.
  3. Always take notes. We live in a world with a wealth of information, and it is important to organise ideas and information.
  4. Do not work in isolation. In other words, do not pigeonhole yourself.
  5. Ask questions early and often; resolving a misunderstanding quickly saves time later.
  6. Understanding is key. Some work is largely engineering, but we should still be able to explain the concepts behind it.
  7. Acknowledge when you do not understand something; when we seek knowledge, help and support, we will receive it.

All posts