Artifacts

Here are some projects I’ve worked on, plus a few things I’ve collected along the way.

TablePage.ai & Revv Research

I’m building tools to make scientific data more accessible and trustworthy. TablePage extracts, structures, and hosts datasets from published research papers. Revv Research accelerates the research process itself.

For my thesis (supervised by Kate Larson at the University of Waterloo), I was interested in modelling users’ behavior on Twitter. I wanted to know: What is the life cycle of a popular tweet? What sort of content do users hesitate to post online?

I also wanted to more generally understand online self-censorship: STAST 2023 Work in Progress: A Glance at Social Media Self-Censorship in North America.

Honors Thesis: Deep RL for Solving Combinatorial Games with Exponential Action Spaces

Awarded the prize for best thesis in computer science. The winning project report and the corresponding slides.
This project is joint work with Ben Burk. For more, check out the codebase.

Abstract: We present a new policy architecture that operates efficiently with a large number of actions. We apply this approach to a family of games described by Erdös and Selfridge to reduce the action space from exponential to linear. We achieved strong performance and demonstrated generalization over a large range of instances. We also explore a generalized variant of the game, and present theoretical results for optimal play against sub-optimal agents.

Convolutional Neural Networks for Image Denoising

In summer 2019, I was at the Shanghai Jiao Tong University prototyping new convolutional neural network models for image denoising. For more, check out the codebase.

Neofields

Some research in theoretical CS (finite fields, Latin squares, covering arrays) for an Undergraduate Research Opportunity Award (UROP).

Egyptian Fraction Multisets

A cute problem that has stuck with me since first year of undergrad. An Egyptian fraction is any fraction that is 1 divided by a positive integer. I am looking for multisets of these fraction that satisfy:

The sum of all elements in the multiset is an integer.
No subset of the multiset adds to exactly one.

The minimal example: ${ \frac{1}{2}, \frac{1}{3}, \frac{1}{3}, \frac{1}{5}, \frac{1}{5}, \frac{1}{5}, \frac{1}{5}, \frac{1}{30} }$

Unproven Claim: There does not exist such a multiset with sum $N$, without the product of its denominators containing at least $N+1$ distinct primes. (In the previous example, the sum was 2, while the denominators contained 3 primes: 2, 3, 5)

Look into Engineer Without Borders’ Junior Fellowship Program! (I was president of the uOttawa Chapter - chat with me if you are thinking of doing it).

I really enjoyed my time at the University of Washington through the Killam Fellowship.

Here’s my valedictorian speech from graduation.

Favourite Blogs:

Favourite Painting:

Favourite Quote: “The Tzadik (the righteous one) is he who rediscovers in adulthood what he lost in childhood.”

Favorite Movie: Perfect Days

Wei Hu