NCTU Observations and Reflections (3): First-Year Master's (Spring) — Fine Arts Club & Research

¶ Preface

Following the previous post, which covered my coursework in the spring semester of my first year, this time I will talk about extracurricular activities and research. My only extracurricular was the Fine Arts Club. Compared with my undergraduate years—interning everywhere, contributing to open-source projects, and participating in communities—this semester felt much simpler. Because I decided to take my master’s program seriously relatively late, I was already behind when it came to exploring directions and finding a topic. Still, this semester I found a research area that genuinely interests me and also settled on a topic, and I began working on it over the summer.

¶ Fine Arts Club

I also mentioned the Fine Arts Club in the M1 fall post. In short, I kept going to the club—actually, I continued into the next semester as well.

The main event in the second half of the semester was a joint exhibition of arts-and-culture clubs in early April, where we set up an exhibition space in the hall beneath the library. This year, because of COVID-19, the semester started late, which squeezed the creation timeline. We had to finish everything before the end of March, so everyone was busy trying to complete their pieces as quickly as possible.

At the time I did not know what to paint, so I decided to paint a starry sky. I originally wanted something dazzling, but as I painted it somehow drifted into an abstract style. It feels a bit like a nod to Van Gogh, though the technique is actually quite different. Van Gogh applied paint very thickly, which makes the colors look rich and saturated. Oil paint is expensive, and I did not want to waste it, so I only applied a thin layer; the result is a brighter-looking canvas.

Overall, I was pretty satisfied with the result. Creative work really does take unexpected turns!

(Starry Sky. Liu An-Chi. 12P oil painting. 2020)

After the exhibition, I still felt like I had not had enough, so I started the next oil painting. When I was little, I once saw a huge oil painting at my uncle’s house: a black background with several slender white tree trunks. Inspired by that, I created this piece, “White Forest.”

(White Forest. Liu An-Chi. 20P oil painting. 2020)

Oil paint takes a long time to dry, so when painting you typically apply one layer at a time and then wait two or three days for it to dry before continuing. A single piece often ends up with many layers. This one, however, took only three sessions from start to finish. The first session was the base coat—painting the entire canvas black—which still took quite a while (about an hour) to cover everything.

The second session was painting the pure-white tree trunks in the foreground. I actually used my fingers rather than brushes, because I found finger-painting much more natural for this. The third session was the background: grayish-white branches that mainly serve as supporting elements, and I also used my fingers for those.

I started by looking for a forest photo and painting from it. The foreground is mostly painted from reference, but later I painted the background by feel. The goal was to keep the overall composition harmonious while intentionally breaking it in small ways—for example, there is a long branch in the background that extends from the left toward the lower right and almost spans the entire painting, creating a deliberate sense of dissonance. In addition, there is only a single branch in the upper middle that stretches upward flamboyantly; I left most of the surrounding area black to emphasize its unruliness.

Strictly speaking, this canvas is about half a person’s height, but compared with my uncle’s painting (roughly two to three meters) it is still much smaller. Later, when I placed it at home, it felt not “grand” enough—but it is still my favorite piece so far, and I even set it as my phone lock screen.

The photo below shows the instructor demonstrating a portrait during a pastel session.

And these are the various pastel drawings everyone made afterwards 😂

In one class, everyone played with paper clay.

We also had a joint club class with the ceramics club, where we painted on cups and shaped a ceramic bowl.

¶ Research

Back in undergrad, I wanted to do a project with a professor. At the time I was deeply interested in browsers. I looked around the NTU CS department and felt that Professor 洪士灝’s research was the best fit, so I ended up approaching him more or less by accident. Later, I also did an MOST undergraduate research project with him. He asked whether I wanted to work on GPUs, but I was completely obsessed with browsers back then—I only wanted to study browsers.

It was not until midterms last semester that I started looking for a lab. I did not really have many options, and by coincidence I joined Professor 游逸平’s Systems Software Lab. I started by reading papers somewhat randomly, and I went through several OpenCL-related papers. They felt interesting, and I began to grow interested in heterogeneous computing and parallel computing.

Later, I attended a meeting at Professor 洪士灝’s lab and heard about something new to me—RDMA. It was eye-opening. That was when I began learning about the field of high-performance computing (HPC).

HPC research equipment is usually expensive. Without money, you cannot really play with it. But I was very interested in this area, which honestly made me anxious. I desperately wanted to know how to train myself into an HPC expert. Once, I went back and asked Professor 洪士灝 for advice. He did not seem to have an easy way to explain it, but later I understood why: the best way is to experience it firsthand.

He suggested that if I was interested, I could apply for an internship at the National Center for High-performance Computing (NCHC). For me, that would be an incredible opportunity—if I could go to NCHC, I could dive into HPC as much as I wanted. At that time, I had already chosen my master’s thesis topic: offloading for parallel web computing. It should also be possible to connect it with the research at NCHC.

The process of finding my thesis topic was also somewhat accidental. One day my classmate 陳奕安 said there was a paper he found interesting, so I asked him for it. After reading it, I felt reproducing the results would not be difficult, and there were many aspects that could be improved. I looked up related papers and realized no one had thought about certain angles—so I found my topic. Because I was taking four courses, I did not have time to properly start, so real progress only began after the summer started. When my thesis is done, I will write a separate post explaining the whole story behind my research.

It is also funny: my undergraduate project idea was browser optimization—modifying the browser to render pages faster—which was more like working in kernel space. That attempt went nowhere because I was too inexperienced. Yet in the end I came back to a similar domain. This time I also want to optimize web rendering speed, but the work moved to user space: modifying Web Worker and WebAssembly.

I read quite a few papers this semester, and I could clearly feel the joy of learning. Knowledge really accumulates like this: cutting-edge technology is published as papers; classic papers become textbooks; and classic textbooks become widely taught and passed on. After reading enough papers, I finally started to feel like I was actually exploring a discipline. It also feels different from reading papers in undergrad—perhaps because I can connect knowledge across areas more effectively now, and I have a sharper eye for the problems.

I hope to complete my research as soon as possible and make it solid enough to publish at a conference. I am not sure whether I will want to pursue a PhD in the future, but having at least one conference paper would give me something to stand on if I ever decide to apply. Having a PhD degree does sound pretty badass, right?! 🤣

¶ Conclusion

This post records my activities in the NCTU Fine Arts Club and my research progress during the spring semester of my first year in the Institute of Computer Science and Engineering. I created quite a few works in the club and also found my research direction. Hopefully I can complete the research smoothly.