Updated on July 16, 2020.
Today’s meeting with my supervisor inspired me to re-think where we should submit and publish our researches. Mainly, pay careful attention to the scope and aim of journals. And, at the very beginning of writing up a paper, i.e. a framework level, think and decide where to submit. This can affect how you write and what information to include. A good example is IEEE Robotics & Automation Magazine (RAM). Its
Aim and Scope is
” The IEEE Robotics & Automation Magazine is a unique technology publication which is peer-reviewed, readable and substantive. The Magazine is a forum for articles which fall between the academic and theoretical orientation of scholarly journals and vendor sponsored trade publications.
The IEEE Transactions on Robotics and the IEEE Transactions on Automation Science and Engineering publish advances in theory and experiment that underpin the science of robotics and automation. The Magazine complements these publications and seeks to present new scientific results to the practicing engineer through a focus on working systems and emphasizing creative solutions to real-world problems and highlighting implementation details.
The Magazine publishes regular technical articles that undergo a peer review process overseen by the Magazine’s associate editors; special issues on important and emerging topics in which all articles are fully reviewed but managed by guest editors; tutorial articles written by leading experts in their field; and regular columns on topics including education, industry news, IEEE RAS news, technical and regional activity and a calendar of events. “
The key information about RAM are:
- Similar to JFR, it focuses on working systems and solutions to real-world problems, verification on real robots are essential.
- Good tutorials articles are welcomed.
I just finished reviewing a paper submitted to a top journal on May 9, 2020. Here are some lessons I learned from it about writing my own paper:
Do NOT afraid to submit your paper to a respectable journal, it’s NOT that hard to get accepted. Even got rejected, you get useful reviews and increase the possibility to be accepted later.
Write logically and concisely. Think what’s the purpose of each section, paragraph, and sentence. Don’t add meaningless sentences. For paper length, technical details are way more useful information.
Use pictures well. Do give clear explanations under each picture. Pictures are key to make a paper easy to follow and well organized.
Don’t bluff, don’t talk big. Speak honestly and clearly what is your contributions. Engineering works are great contributions. Interdispline is not necessary unless it’s something.
After going through the book How to be a Modern Scientist by Dr. Jeff Leek, I am very impressed by his precise and manageable methods to gain more success in academia. Especially his academic paper writing deeply speaks to me. Here are the most highlight points I will follow for all my future paper writings.
- When to start writing?
- One good principle to keep in mind is “the perfect is the enemy of the very good” Another one is that a published paper in a respectable journal beats a paper you just never submit because you want to get it into the “best” journal.
How to start writing?
- Once you have a set of results and are ready to start writing up the paper the first thing is not to write. The first thing you should do is create a set of 1-4 publication-quality plots (see Chapter 10 here). Show these to Jeff to get confirmation on them before proceeding.
- Start a project on Overleaf and invite Jeff to join.
- Write up a story around the four plots in the simplest language you feel you can get away with, while still reporting all of the technical details that you can.
- Go back and add references in only after you have finished the whole first draft.
- Add in additional technical detail in the supplementary material if you need it.
- Write up a reproducible version of your code that returns exactly the same numbers/figures in your paper with no input parameters needed.
This is golden advice. I also found the paper outline he recommended is super helpful to make the paper organized and easy to check what part you are missing. I quote as my personal notes as follows.
Abstract is structured like introduction, but shorter, typically 125 to 250 words, definitely only 1 paragraph.
The introduction has a minimum of 3 paragraph, possibly more, all following the OCARF structure.
Opportunity: This is described in a single paragraph. Convey that there is now some awesome opportunity emerging, because our colleagues have been working very hard doing brilliant stuff. We are on the cusp of learning something fundamental, that we’ve never been able to ascertain before. Remember to funnel down, from most general opportunity (eg, understanding the neural code for behavior), to the very specific one (eg, understanding which neurons may be causally involved in larval drosophila behavior). Also remember that this is an opportunity to tell your readers (upon whose shoulders your work is built), how awesome they are. They did so much great work to get us to the point that we could add this cherry on top.
Gap: Last sentence of 1st paragraph: but, there is a gap: something that we don’t know, how don’t know how to do, that resolves the age old mystery.
Challenge: At least 1 paragraph, maybe 1 per challenge. Explain that resolving this gap, to address opportunity, is very challenging for a number of reasons. State reasons from most obvious to least. For each, give credit to colleagues who have brought us much closer, though not close enough yet. This is another opportunity to remind the readers how great they are, and how you could not possibly have done with work without standing on their shoulders.
Action: This gets about 1 paragraph, maybe a little less. Key is explaining how the action we took addresses the challenges we faced, to fill the gap that existed in the literature, which now resolves the original opportunity.
Resolution: This couple sentences resolves the gap. In other words, there was an opportunity, that opportunity no longer exists. This likely includes: (i) theory, (ii) simulated experiments, (iii) real data analysis, (iv) model checking / synthetic data analysis.
Future: 1-2 sentences explaining what new opportunities now arise, by virtue of our resolution.
First result always illustrates the main conclusion from the paper, either via a simulation or real data example.
Next is a toy problem, that enables us to build our intuition as to why this approach is useful.
Stress test the method, demonstrating the extreme cases of where it works when it should, and even when it shouldn’t. Possibly this includes simulated benchmarks.
Some real data analysis, demonstrating that our approach outperforms other methods on several benchmark datasets.
Experiments on Novel Datasets
A motivating example perhaps, that justifies the development of this method in the first place.
Summary: In a paragraph, summarize, in a reverse funnel fashion, stating the precise result, and then zooming out to show its relationship to the more general problem. Explain how this result changes the life of the reader. What can she do now, that she couldn’t do before?
Related Work: In about 2 paragraphs, talk about the most closely related results (~1 paragraph per). Remember, the authors of these works are your reviewers, colleagues, and friends, so be as generous as possible, without stomping on your own parade. There is no reason to be at all negative of anybody else’s work, just highlight the advantages of this work.
Future: Explain how this work enables the next work. What problems are now open, that we not possible to address before, or perhaps not even conceived of before? This also should funnel out, from the most specific problems, ending with the most general ones.
These are articles [@Caruana08], [@Delgado14].
Problem: In a paragraph, describe the problem. Start with the most general formulation of the methods, and funnel down, much like in the first paragraph of the intro. For example, perhaps start with supervised learning, then classification, then 2-class classification, then high-dimensional two-class classification.
Model: In a few sentences, write down the most general statistical model under investigation. Perhaps it is non-parameteric, or perhaps it is a state-space model. Then, start making a series of simplifying assumptions. For each assumption, justify why you’ve made it. The only valid justifications that I now of are: (i) analytic tractability, (ii) computational efficiency, (iii) physical realism, (iv) reducing variance, or (v) increasing interpretability/understandability.
Algorithm: In a paragraph, describe the algorithm. Be as specific as possible. Include a pseudocode table, possibly to be put in appendix. Refer to equations in text where possibe.
Simulations: Describe simulation settings. This should essentially be parameters of the model specified above. 1 paragraph/bullet point per simulation study.
Data: In a couple sentences per dataset, describe it. Be sure to include, at a minimum, (i) a link to where it came from, (ii) the number of samples, (iii) the dimensionality, (iv) if any missing data, explain, (v) if any know structure, explain.