Nothing is more practical than a good theory.
All human wisdom is summed up in two words? Wait and hope.



在研究中,我的志趣在於智能和學習 (intelligence & learning)。近年來,基於機器學習所開發的各類智能應用對我們的平凡生活帶來了許多的影響。然而,對於智能本身,一些最基本的問題還遠遠談不上解決。無根之木,難以參天。在未來,我希望能夠通過自己的工作,發掘現有的問題,理解產生這些問題的本質,並據此設計出更好的智能。

Debug 時出現的無聊幻想:如果努力向上,總有一天蓋鳴暉會娶我的!(注:蓋鳴暉小姐是著名粵劇藝術家,在舞台上是文武生,亦可應花旦。我最喜歡的她的幾齣戲是<梁祝之樓台會>和<南唐李後主之去國歸降>。我覺得她的表演,「真謂以血書者;儼有釋迦、基督擔荷人類罪惡之意」。)

Last updated in 2020.

I just wondered how things were put together.
I am very seldom interested in applications. I am more interested in the elegance of a problem. Is it a good problem, an interesting problem?
My fondest dream is to someday build a machine that really thinks, learns, communicates with humans and manipulate its environment in a fairly sophisticated way.
Claude E. Shannon

Hey, this is Ziyu, a pursuer of machine intelligence.

I am currently a Ph.D. student in Computer Science at The University of Chicago. Before this, I completed my Master of Science degree of Computational Analysis and Public Policy there. I had my Bachelor of Economics degree at The Special Class for the Gifted Young.

In doing research, I was primarily driven by the curiosity to understand the learning mechanism of intelligence. The development of deep learning applications have influenced the quality of our life greatly in the last decade, however, the theoretical foundations of intelligence is far from established. There are lots of questions needed to be answered, and more questions needed to be asked. My goal is to identify and solve for the problems critical to understand learning mechanisms, and generalize principles useful for us to design better intelligent machines, which really think / learn / communicate with humans, as Shannon imagined a few decades ago. I understand the future work to be done by the 4 steps: to identify, to understand, to abstract, and to generalize.

Besides this, I am interested in economics, which provides powerful analytical tools in understanding outcomes by intelligent agents. Economics helps design high-level policies for real-world living problems, which improve social welfare. While I am studying the highly abstract problems as mentioned previously, I often, irrelevantly, feel a sense of guilty for those human beings suffering from mundane life problems (e.g. poverty, political oppression, etc.). I’ve spent some time in rural mountain areas and observed the difficulty for some people to “live” with dignity. In future, I hope I could also contribute my efforts to those problems and those people, instead of just staying comfortably with abstract machine thoughts.

I am a fan of Ellen Loo and Joyce Koi.

Last updated in 2020.