I present here 101 research lessons from my advisor, Professor Young-Koo Lee, which I wrote down during research meetings, seminars, presentations, and discussions over the years. I include them here for the benefit of graduate students as practical resources for surviving graduate school, conducting research, writing papers, and communicating ideas clearly. Together, they reflect lessons that shaped my understanding of research as both a technical and communicative practice, and they have served as invaluable guidance throughout my graduate journey.
— Numan Khan
Terminology, Definitions, and Clarity
1Choose your terms very carefully.Dec. 2, 2025
2Every notation has its purpose. If you use different notations for the same thing in a paper, readers will become confused.Aug. 31, 2022
3In research, define terms clearly. Some terms may overlap, but if there is overlap, you must clearly distinguish them.Sep. 19, 2024
4Use well-defined words in papers. Do not invent your own meanings or use your own interpretations of popular terms. Use terms whose meanings are already shared and well understood.June 21, 2024
5Avoid using abbreviations in papers. Use them only for frequently used terms.Sep. 15, 2022
6"Very" is an ambiguous word. It should be measurable. For example, instead of saying "very costly," explain how costly it is.Apr. 23, 2023
7"In progress" is a very bad term. Even 0.1% complete is technically "in progress." When you say "in progress," it often means less than 1%.May 4, 2023
8Try to use conceptualized terms.Jan. 17, 2023
9Good papers create good names for new concepts.June 28, 2023
Research Mindset and Originality
10One paper should have one idea.Apr. 29, 2024
11Since this is research, you need new motivations. Do not simply follow existing motivations.June 28, 2023
12Research is self-driven work. Do not borrow ideas from the literature. Efficiency should not only be in implementation, but also in research.July 22, 2022
13To be a good researcher, you must know how to throw away less important things.June 21, 2024
14As a researcher, you must have curiosity.July 13, 2023
15You should be the god of that world. You should be the creator. If you understand everything in that world, you can change it however you like.July 13, 2023
16In research, we have to know why. Software tells you how to use something, and you followed it, but you did not think about why. For research, we must understand why.Aug. 31, 2022
17You have to think first and then validate your thinking through experiments.Feb. 12, 2023
18If your method gives a 2× improvement, ask why it is 2×. Why not 3×? That is research.Mar. 26, 2023
19In research, not every attempt can succeed, so we have to try many times.
20What are the strong points? What is unique? What can you advertise? Find truly strong points that people can recognize and accept.
Problem Framing, Motivation, and Contribution
21For presentations, start with motivation and intuition. Clearly explain the current status, the drawbacks, and then the intuition.Apr. 22, 2024
22These days, people think AI and deep learning are the solution. But what problem are they solving? Give the problem first. What is missing?Apr. 22, 2024
23Think about what first, not how. When the problem is clearly defined, anyone can propose a good solution. For example, saying "I am looking for a good wife, and I will use deep learning" is meaningless unless you first define the properties of a good wife.Apr. 22, 2024
24Give the motivation first, then the intuition and key idea.July 15, 2024
25Discuss the drawbacks, provide intuition for solving them, and then present the achievements.July 1, 2024
26First explain the main idea of the related work, then present the motivation. Explain how your work is different from existing work, and then present your main idea in detail.May 8, 2023
27New contributions are the key in research.May 8, 2023
28Contribution means how unique the idea is and how large the impact, performance gain, or improvement is.June 21, 2024
29When you write a paper, you are selling your paper. If you do not emphasize its selling points yourself, readers will not know what they are. You have to state them explicitly and in the right way.Sep. 15, 2022
30Use your time to devise new methods and algorithms.Apr. 30, 2023
31Your work should improve through design, not merely through fine-tuning the implementation. Otherwise, it will not be your contribution.Apr. 30, 2023
Experiments, Evidence, and Analysis
32Without analysis, an experiment is not worth anything.July 13, 2023
33Showing good performance improvement is good, but without analysis, nobody will buy it.May 26, 2026
34Do experiments to find out why your method improves, not just how much it improves. "Why" is the key.July 13, 2023
35You should be the god of the sampling and training world. Do not stay outside; you have to dive inside.July 13, 2023
36When you understand the "why," then you can say you have done research, and then you can write good papers.July 13, 2023
37If the reasons are not supported, then the experiment is garbage.July 9, 2023
38State the claim first, then provide supporting items such as charts.May 8, 2023
39Give a reason for everything. If something is good, explain why. If something is bad, explain why.May 8, 2023
40What supports your claim? You have to support your claim with algorithms, data structures, and complexity.
41GB and GiB are different. If you see something different, you have to find out why.June 4, 2023
42For charts, if there is more than one type of data, use two vertical axes.June 4, 2023
43Use self-describing captions for all charts.Sep. 16, 2022
Technical Writing and Paper Organization
44Technical writing is the opposite of detective novels. Detective novels create suspense. Technical writing explains things so readers can understand easily.July 15, 2024
45Wording is important. Even if the research result is the same, its evaluation can be very different depending on how you describe and present it.Sep. 15, 2022
46Papers are collections of justifications, not facts.July 15, 2024
47Do not mix logic and physical things.July 15, 2024
48Do not show the blueprint and the physical building together.July 15, 2024
49In a paper, each paragraph should have its own subtopic.July 1, 2024
50The first sentence is the key sentence.July 1, 2024
51The key sentence provides a map for the readers so they are ready for the following content.July 1, 2024
52Structure and organization are the most important.July 1, 2024
53Do your homework yourself. Do not make the readers do your homework.July 1, 2024
54Paper writing should be like object-oriented programming, not procedural programming.July 1, 2024
55When writing papers, write in a top-down manner so readers can predict what comes next.Nov. 27, 2022
56Read a paper once for its technical ideas, but read it once more to learn how to write and present well.June 28, 2023
Presentation and Slide Design
57A presentation is not about repeating the history of your study. It is about writing a new novel or story.Sep. 10, 2024
58In a PPT, each page or slide should have its own subtopic. The header is the main topic, and the bullets support the header.July 1, 2024
59The goal is not to show the steps. The goal is to show effectiveness.June 21, 2024
60Code can be shown as a reference, but it should not be the major part of the presentation.June 21, 2024
61Do not explain with code. Explain the concepts first. We are not machines.July 23, 2023
62Code is for machines, not human beings. Explanation by code is not desirable.Aug. 28, 2023
63Present the key idea first before going into the details.May 8, 2023
64Being able to explain without a figure is more important.Jan. 17, 2023
65You are like painters. You let visitors generate their explanations from your paintings.Jan. 17, 2023
66Briefly express the idea in text, and then let the figure confirm it.Jan. 17, 2023
67Make sure you deliver your ideas using text so that readers get the correct understanding of your work. Then, illustrations should confirm that idea.Jan. 17, 2023
68The best representation is to use one title per slide. If you present sequentially, you do not need extra context.
69A slide should be understood quickly. Usually, it should have one or two lines followed by itemized points. Do not make a slide a copy of the paper.
70A presentation is a service to the audience. Make sure you do justice to the audience.Aug. 13, 2022
71When I ask you something, sometimes I already know the answer, but I am teaching you how to present and how to develop your presentation skills.Aug. 28, 2023
Literature Review and Domain Understanding
72If you research a domain, you must fully understand everything.July 16, 2023
73Authors do not put everything in a paper. It is your job to find the hidden assumptions, meanings, and method behind the paper.
74You are not high school students.July 16, 2023
75Select papers and choose keywords. Then search based on those keywords.Sep. 18, 2022
76Create a multidimensional space. Put each paper in one dimension. Then you can find new subspaces.Sep. 4, 2023
77To break the literature trap, specify your goal and take time into consideration. Break the loop and move to implementation.Feb. 7, 2023
78If you do not have content to present, then you should present recent papers.July 13, 2023
Time Management, Discipline, and Work Habits
79Please do not waste your time.Apr. 29, 2024
80If you make a loose schedule, it will lead to bad habits.Feb. 15, 2024
81Study how to use your time efficiently. You do not have endless time.Sep. 4, 2023
82Do not try to find excuses for delaying your work. Follow a fixed timeline. Do not be generous with yourself.Sep. 4, 2023
83Rather than postponing the schedule, finalize things one week earlier. Ask yourself: what should you achieve by the end of this week?Feb. 7, 2023
84You have to estimate what you need to produce in the middle of the process. Estimation is one of your capabilities.Feb. 7, 2023
85Estimation requires knowing the system well. You must know every component in the system. Estimate the time and predict how long the work will take. Evaluate your prediction. If it was wrong, analyze why it was wrong.Feb. 7, 2023
86When you have worked hard, you have many things to show.Nov. 13, 2023
87You are the researchers, and I am just the advisor. Do not make me push you; that is the wrong approach.June 16, 2023
88You should push yourself to the extreme to bring out your hidden capability.Nov. 24, 2022
89Think about your competitors around the world. Do you think they all enjoy festivals? With this lifestyle, do you think you can compete with them?Apr. 20, 2023
90The graduate study period is the hardest period of your life.Apr. 20, 2023
91You have to conduct research by enjoying the research.Apr. 20, 2023
92Before reaching the core joy of research, you will have trouble.Apr. 20, 2023
93The other person's cake looks bigger. (The Korean saying is 남의 떡이 더 커 보인다.)Sep. 4, 2023
Communication, Feedback, and Asking Questions
94Research is communication.June 6, 2023
95Learn how to ask.May 29, 2023
96This is not the place where you present only answers. This is the place where you show issues and approaches and receive feedback.Nov. 14, 2022
97Ask why about everything.Nov. 27, 2022
98If you cannot find help by yourself, then finding help is your skill.Nov. 24, 2022
Implementation, Design, and Frameworks
99Study how to draw frameworks and architectures.May 11, 2023
100If there is any issue in the implementation, take notes and discuss it. It can be expanded into new ideas.Apr. 23, 2023
101We are not handling toys. We are handling huge graphs, so existing toy methods will not work for large graphs.Aug. 13, 2022