It is my second time submitting a VIS paper. This year I visited NYU and coauthored the paper with Prof. Bertini. It’s really exciting to work with Enrico and we are satisfied with what we have done so far. Hope there would be other opportunities to collaborate with him. Though it’s not my first time working on the VIS deadline, this is still a fresh experience and I learned a lot from it.
Life in New York
New York is not very different from Hong Kong or Shenzhen, except its horrible subway system. During the stay in NYU, I lived in Jersey City, where I can afford for a good and private room. The neighborhood is actually very nice. There is also a Chinese supermarket in walking distance, where I can buy food and deserts that I enjoy.
It takes about 40 minutes from home to the lab. Every morning, I take a PATH train to World Trade Center, where I then transit for MTA. During the day, the WTC is full of people, some of which are rushing for work, and others are traveling around taking photos. It’s also an interesting experience to see how empty it is in 3 a.m.
This time I did the whole paper with Prof. Enrico and Prof. Qu. At the beginning, I thought it would be a good practice of my research skills if I did the whole paper by myself. However, through this experience, I realized how important teamwork and collaborations are. If I could do the submission another time, I would rather have some other students working together.
It’s annoying to switch my mind from coding to writing and then back. It’s depressing when the project gets stuck and the meetings with professors are three days later. I was missing how I can discuss freely with someone any time things go wrong.
A trick I learned to overcome this negative mind is, when stuck, put down the work in hands and go out for a walk. And sometimes it turns out to be just a minor technical issue that we can bypass. 当局者迷，旁观者清。
Though my skills in writing, coding, and idea generations greatly improved after this submission, it’s still a pity that my social skills and teamwork skills did not get the same improvements.
Research in Explainable AI
It’s also great that I can dig the field of explainable machine learning further using visualization. During my PQE, Prof. Qu raised a point: selling visualizations to ML/DL experts is hard; while selling it to managers is easy. I think this is a valid point for XAI research, and this is the point that drives my submission this year.
The target users should not be limited to those who develop the system and the models. There is a significant demand to help people without knowledge in ML to understand/trust/use ML techniques. I think this point also accords to part of the goals of massive online education.
Automated solutions powered by ML will be increasingly adopted in our daily life. How the algorithms make things right/wrong? Why? There will always be people interested in these questions, and VIS would be a critical and effective step towards the answers.
I always enjoy making beautiful graphics. I think this is one of my driving powers for research in VIS. Apart from research, I felt a great sense of achievement by making them and having people like them.
From the past five years (I have been making posters for different student organizations during my undergraduate), I learned that I don’t have much talent in design or art. But making graphics can be count as one of my habits. I would not feel tired of creating graphics for a whole day (but would feel tired of writing). I think it’s very lucky for me to be able to join Prof. Qu’s group to work on VIS.
By my own standards, my skill sets in coding improved a lot via this submission.
I learned how to write Python C extensions to glue code written in C for performance. I learned React and Typescript in front-end developments. I learned Docker when I need to set up demos to run user studies.
Another lesson I learned is, don’t change the framework or refactor the code if the time is limited. During the submission, I wasted about 5 days, changing the code from D3.js to pure React (use React to handle the rendering of SVG nodes) and then back to D3.js. When efficiency is crucial, it’s worth sacrificing the readability gained from React.
To build things with my own code is one of the most exciting things that I love in Computer Science.
Things I learned from Enrico and his students
I learned a lot from Enrico during the collaboration. Enrico has his standards for what is good research. Though he is quite open to my own ideas, his standards helped me a lot in refining the direction and scope of this paper. I think a good measure for the value of an application research can be roughly formulated as: the potential value added to a user x the number of potential users. Prof. Qu also suggest that we should working on more important problems (而不应该去研究“茴”的四种写法).
One important thing I learn from him is that, don’t presume that something is obvious. As researchers, we work on our projects every day. Something clear to us might not be easy-to-understand to others. It’s always He tried to understand the most detailed parts of my proposed method and pointed out things that he thought are fragile. This helps a lot for this submission.
Another thing I learned from his group members is to stay critical. During meetings, they always actively raised points they felt confused or ideas they felt relevant. They would start from a question raised from a presentation and discuss freely for half an hour. Presenting our work and discuss it with people is the quickest way to improve it. It turns out these discussions are always valuable and would contribute to one’s submission.