Before he headed to Oklahoma City to attend the 2022 IEEE VIS conference, we chatted with Hyung-Kwon to ask him questions about his journey in choosing a career in human-computer interaction (HCI), internships, working on projects that interest him, and how he learned to prioritize his health.
Dive into the insights in this interview with Hyung-Kwon as he shares his journey to becoming a published researcher.
How did your undergraduate education in mathematics prepare you to pursue a graduate degree in computer science?
Hyung-Kwon Ko It was a series of choices that led me to study computer science. Although studying mathematics was interesting, I was not sure how studying math could help solve important real-world problems. Thinking about the practical values of studying a major as an undergraduate student, I started studying industrial engineering and statistics, as they seemed to deal with more down-to-earth problems. After taking several courses, I realized that industrial engineering covers diverse lectures from economics to computer science. I found myself fitting best with the courses like Data Mining, Data Structure, and Algorithms.
Although it was a bit late to study computer science since I was a senior then, I registered for an additional semester. I took a total of 16 credits with students majoring in computer science. I had to struggle as I did not have pertinent background knowledge compared to those who studied computer science as their major. It was a painful time, but I grew greatly through that experience. Since I wanted to study more, I decided to go to graduate school.
What got you interested in HCI?
Hyung-Kwon Ko When I took Data Mining as a class, I learned what principal component analysis is. It was a famous dimensionality reduction technique, and instantly it seemed fascinating. It enabled compressing the size of the dataset with a minimum loss of information. Frankly speaking, I did not know what HCI was or what it did, but I looked for labs researching dimensionality reduction techniques.
After a few months of searching, I found a lab at Seoul National University and realized that the field I was looking for was called HCI (Visualization). I emailed a senior graduate student to check whether what I was looking for was correct, and he confirmed that I was right. I joined the lab as an intern and kept researching HCI and visualization.
What was your process in looking for an internship? Were internships introduced to you by advisors, or did you have to look independently?
Hyung-Kwon Ko It was done all by myself. Although I planned to apply for a Ph.D. right after graduation, I thought I was not ready as I did not have enough publications. Therefore, I started finding places where I could go for further research experience. At that moment, I had to think about myself. What do I like the most? This was an important question as I had to choose the domain I was applying for my future career.
I narrowed down the domains into a few and finally chose webtoons as I really like them. As I was taking coursework in graduate school that required a final team project without a specific topic, I decided to develop a software program for webtoon authors. However, I had no idea what their domain problems were.
I cold-contacted more than 15 professional webtoon authors as I did not have any acquaintances, and fortunately, I could have interviews with a few of them. By having consecutive interviews with them, I learned important domain problems that need to be solved. Specifically, the communication between the writers and artists was inefficient, mainly because the writers did not have proficient drawing skills. Thus, we proposed an interactive system where writers can convey their revision intention with simple clicks by modifying any user-given images.
Although it was a project for the coursework, I wrote it down on my resume and had a chance to work at Naver Webtoon Corp. as a research intern.
What steps did you take to transition from an intern to full-time researcher?
Hyung-Kwon Ko Since the program, I developed was a simple demo version, I had to develop it as a more sophisticated prototype. In this process, I used real images by collecting more than 47K images from 441 webtoons. I trained state-of-the-art generative models using these datasets and integrated them with the front-end user interface. I also had to recruit 24 professional webtoon authors for evaluation, and the result said the system helped the communication to be more efficient and effective. The work was also published as a conference paper in ACM UIST 2022. Acknowledged by my commitment to this work, I was able to change my position from a research intern to a research scientist.
Based on your experience, what actionable advice can you share with others ready to become full-time researchers?
Hyung-Kwon Ko When I was doing the research, I was sheerly enjoying the moment. Because I found the topic after having several discussions with professional webtoon authors, I had confidence that solving this problem could make important progress in the webtoon domain.
It is a little weird to say, but I did not think about my position and its transition. I talked and discussed the problem with senior researchers without considering their positions. This was possible because I loved the domain more than anyone. I have more than 400 comic books in my house, and I have read webtoons since elementary school.
Therefore, I think choosing an important problem in one’s favorite domain is critical. I believe other issues will become relatively trivial if this is settled well. Once the direction is decided, the next thing to do is run hard.
What were your biggest challenges in publishing your first manuscripts, and how did you work through them?
Hyung-Kwon Ko I think it is very helpful to have many good senior collaborators around you. At first, I did not know what to do or where to start researching. I asked my advisor and senior graduate students many questions, and they gave me a lot of advice regarding setting research questions, doing experiments, and writing manuscripts.
Before getting advice, my writing style was pointed out to me many times. As I have never tried scientific writing before, I was familiar with essay writing, which frequently uses ambiguous words and sentences. However, they were not acceptable in writing scientific papers.
Following the advice, I read many famous publications, for example, papers nominated for the best paper award in major conferences, to refer to their writing style. Moreover, I took a writing class held for undergraduate students to practice my writing. Although it was still insufficient to write a paper, my advisor and the senior students proofread my paper and helped refine the manuscript. Thanks to their help, I could successfully submit my first paper to the conference.
What advice can you offer to researchers who are nervous about publishing a manuscript because they aren’t native English speakers?
Hyung-Kwon Ko Fortunately, there exists a certain structure in writing scientific papers. I believe that novice researchers can learn the types of formulas in writing each section. To this end, the first thing they need to do is to read many great published research papers. A repetitive reading will teach them how to deal with it. Once the reading is finished, they can start writing anything in English rather than translating from their mother language. They have to be familiar with thinking in English. There is no other way I can say that practice makes perfect.
Pursuing an advanced degree can be demanding. What advice can you offer PhD students struggling to manage stress or other mental health concerns?
Hyung-Kwon Ko Last year, I was in a very stressful situation, which was to submit a conference paper and finish my final intern project simultaneously. I had to work from 5 am to 12 am every day. My paper was rejected, and due to the stress, I had to call 911 twice because of a heart attack and severe dizziness.
Although I have never questioned my health, it happened to me suddenly. After that experience, I realized that taking care of myself is more important than anything in this world.
Since then, I’ve tried to take a break whenever I think it is needed. Now I think it is okay to postpone submitting publications and try to calm myself by walking when I am stressed with a pile of work to do. It does not mean that I work less enthusiastically; rather, I try to be relaxed and comfortable mentally. Also, I go swimming every day and eat meals on time. I understand many Ph.D. students are busy and often in very stressful situations. Still, I believe it is more important to think globally by understanding that I am not the only one as there are my family and friends who support me, and everything will be okay.
What fun, side projects are you working on that are helping you move toward your goal of starting a company with products that innovate people’s working paradigm?
Hyung-Kwon Ko Although it is not a project, I try to meet people in diverse domains as much as possible. I have tried to change my environment to date. For example, I went to three distinct universities and studied three majors (i.e., mathematics, industrial engineering, and computer science). Moreover, I worked for companies in different fields (i.e., manufacturing, IT, consulting). Every time I changed my environment, I met new people with different backgrounds and specialties. This experience made me connect with people in varying domains, and realize where my position is located more globally. At the same time, I could seek important problems waiting to be solved by talking to these domain experts. Moreover, I believe it is not an individual, but a great team that can make great products. After building a healthy and strong network, I will choose one of the items that matter to all humankind and will make a profit out of it.
More About Hyung-Kwon Ko
Hyung-Kwon Ko is a researcher at the KAIST Interaction Lab, a human-computer interaction research group at the Korea Advanced Institute of Science and Technology (KAIST). Tasked with innovating the way we interact and collaborate, Hyung-Kwon designs, prototypes, and evaluates interactive systems that have the potential to shift people’s working paradigms. His primary focus areas include human-computer interaction, human-centered AI, and information visualization.
Hyeon Jeon*, Hyung-Kwon Ko*, Soohyun Lee, Jaemin Jo, Jinwook Seo, “Uniform Manifold Approximation with Two-phase Optimization,” Proceedings of IEEE VIS 2022 (Short), to appear.
Hyung-Kwon Ko*, Subin An*, Gwanmo Park, Seung Kwon Kim, Daesik Kim, Bohyoung Kim, Jaemin Jo, Jinwook Seo, “We-toon: A Communication Support System between Writers and Artists in Collaborative Webtoon Sketch Revision,” Proceedings of the 35th ACM User Interface Software and Technology Symposium (UIST 2022), to appear.