Liang Ze Wong, PhD

I am a mathematician, machine learning research scientist, computational social scientist, data scientist, and full-stack programmer.

From 2019 - 2024, I was a research scientist at the Institute of High Performance Computing in Singapore, working primarily on using natural language processing (NLP) for social media analytics, stance detection, and topic modelling.

In Fall 2022 and 2023, I also taught the Foundations of Business Analytics core course for Masters of Science in Business Analytics (MSBA) students as an Adjunct Assistant Professor at the NUS Business School.

I left IHPC at the end of 2024 to spend more time on AI/NLP/LLM research. I am currently open to work and am seeking research scientist positions in interpretability, alignment and theory of language models.

AI/ML papers

Since the start of 2025, I have written the following short papers:

• Paying attention to facts: quantifying the knowledge capacity of attention layers
  [arXiv:2502.05076]

• ‘Generalization is hallucination’ through the lens of tensor completions
  [arXiv:2502.17305]

I am writing a series of blog posts about these papers on this site. The first one is here.

Back in 2014, I developed a sparse auto-encoder that could handle missing data:

• Imputing Missing Values in Sensor Networks using Sparse Data Representations
  with Huiling Chen, Shaowei Lin and Daniel C.L. Chen
  [MSWiM 2014]

Other ML-related papers

• Dependently typed knowledge graphs
  with Zhangsheng Lai, Aik Beng Ng, Simon See and Shaowei Lin
  [arXiv:2003.03785]

• Minimal Schemas for a Category
  [draft] [poster for ACT 2019]

• An Ordinal Potential Function for Network Selection in Heterogeneous Wireless Networks
  with Tony Q.S. Quek and Michael Padilla
  [ICASSP 2014]

Education

• 2014 - 2019: PhD in Mathematics, University of Washington, Seattle
• 2009 - 2012: BA in Mathematics, Cornell University

My undergraduate and graduate studies were funded by the A*STAR National Science Scholarship.

Computational Social Science research

From 2019-2024, I worked at IHPC with an interdisciplinary group of social scientists, psychologists and engineers led by Joe Simons, on applications of NLP and LLMs to social media analytics, stance detection, and topic modeling.

Social science papers

• Understanding Challenges and Emotions of Informal Caregivers of General Older Adults and People With ADRD: Comparative Study
  with Nova Mengxia Huang, Shirley Ho and Bryan Timothy
  [Journal of Medical Internet Research, 27 (2025)]

• Utilizing LLMs to conduct Thematic Analysis: A Case Study on Focus Groups Transcripts
  with Prasanta Bhattacharya, Brandon Loh, Hui Si Oh, Siti Amelia Juraimi, Nandini Anant, Arunika Pillay, Mary F-F. Chong, Anna Fogel, Florence Sheen and Aimee E. Pink
  [in preparation]

• From traces to measures: LLMs as a tool for psychological measurement
  with Joe Simons, Prasanta Bhattacharya, Brandon Loh and Wei Gao
  [arXiv]

Talks

• Applications of LLMs in Social Sciences, AI Wednesdays (Government-wide Community of Practice for Artificial Intelligence), Singapore (October 2024)

• An Introductory Workshop on using LLMs in Social Science. Summer Institute in Computational Social Science (SICSS), Singapore (June 2024).

Mathematics research

In 2019, I received my PhD in Mathematics from the University of Washington, Seattle. My advisor was James Zhang.

My mathematical research interests are in category theory, homotopy theory, and non-commutative algebra. Specifically: Grothendieck fibrations, enriched category theory, Hopf algebras and cubical sets.

Math papers and talks

• Cofibration category of digraphs for path homology
  with Daniel Carranza, Brandon Doherty, Chris Kapulkin, Morgan Opie and Maru Sarazola
  [arXiv] [Algebraic Combinatorics, 7 no. 2 (2024)]

• Cubes with connections from algebraic weak factorization systems
  with Chris Kapulkin
  [slides for Western Topology Seminar][TeX]

• Smash Products for Non-cartesian Internal Prestacks
  [arXiv]

• The Grothendieck construction in enriched, internal and $\infty$-category theory (advised by James Zhang) [PhD thesis] [slides]

• A co-reflection of cubical sets into simplicial sets, with applications to model structures
  with Chris Kapulkin and Zachery Lindsey
  [arXiv] [slides] [New York Journal of Mathematics, 25 (2019), 627-641]

• The enriched Grothendieck construction
  with Jonathan Beardsley
  [arXiv] [slides 1, 2, 3] [Advances in Mathematics, 344 (2019), 234-261]

• The Operadic Nerve, Relative Nerve, and the Grothendieck Construction
  with Jonathan Beardsley
  [arXiv] [slides] [Theory and Applications of Categories, 34 (2019), 349-374]

• Weak Equivalences between Categories of Models of Type Theory
  with Simon Cho, Cory Knapp and Clive Newstead
  [slides]

• Noncommutative geometry of homogenized quantum $\mathfrak{sl}(2,\mathbb{C})$
  with Alex Chirvasitu and S. Paul Smith
  [arXiv] [Pacific Journal of Mathematics 292 (2018), no. 2, 305-354]

• Irredundant Generating Sets of Finite Nilpotent Groups
  [Senior thesis] (advised by R. Keith Dennis)

Math workshops and seminars

I had the privilege of participating in the following meetings:

• MIT Talbot Workshop 2018
• AMS Mathematics Research Community in Homotopy Type Theory
• Kan Extension Seminar II [blog post, long slides, short slides]
• Budapest Semesters in Mathematics, Fall 2011