Sean (Shouheng) Li

Sean (Shouheng) Li

My research and work.

About me

Excited to announce our paper, “Towards Bridging Generalization and Expressivity of Graph Neural Networks”, will be presented at ICLR 2025 in Singapore this week!

We delve into the connection between expressiveness and generalization of Graph Neural Networks:

  • Graph ML models can be generalizable yet powerful at the same time
  • We offer insights to enhance a model’s generalization pre-training
  • Findings extend to non-graph ML models as well

Catch us at ICLR: 📍 Poster Session 5 #175 🗓 Friday, 25 April 🕙 10:00 am – 12:30 pm SGT 🏢 Hall 3 + Hall 2B

Access the paper here: Paper Link Explore the code: Code Repository

Grateful for the collaboration with co-authors Floris Geerts, Dongwoo Kim, and the late Qing Wang, a remarkable mentor whose legacy lives on.

Looking forward to engaging discussions on ML and graph machine learning at the event!

I am a graduate Ph.D. researcher in the Graph Research Lab of ANU, led by A/Prof Qing Wang and Prof. Brendan McKay. I am also supervised by Dr. Dongwoo Kim from POSTECH.

My research interests center around two main topics: (1) studying the foundations of geometric deep learning, especially the expressivity, generalization and optimization of graph machine learning; (2) Exploring the novel connections between between mathematical theories (e.g. graph theory) and machine learning to help designing more robust, powerful, and explainable models.

In addition to my academic pursuits, I hold the position of Senior Research Engineer at Data61@CSIRO. Prior to commencing my PhD, I had the luck to work with the StellarGraph team, where I contributed to various projects including the open-source graph machine learning libary.

With extensive industry experience in software engineering, I have also worked across diverse domains such as data analysis, video/image processing, and medical care.

Recent Publications

  • ICLR 2025, Towards Bridging Generalization and Expressivity of Graph Neural Networks. Shouheng Li, Floris Geerts, Dongwoo Kim and Qing Wang, paper

  • ICML 2023, Local Vertex Colouring Graph Neural Networks, Shouheng Li, Dongwoo Kim and Qing Wang paper, live demo, slides, poster, code

  • ICLR 2023, N-WL: A New Hierarchy of Expressivity for Graph Neural Networks. Qing Wang, Dillon Chen, Asiri Wijesinghe, Shouheng Li, Muhammad Farhan paper, slides, poster, code

  • AAAI 2023, Restructuring Graphs for Higher Homophily via Adaptive Spectral Clustering. (oral), Shouheng Li, and Dongwoo Kim and Qing Wang paper, slides, poster, code

  • ECML-PKDD 2021, Beyond Low-Pass Filters: Adaptive Feature Propagation on Graphs. Shouheng Li, Dongwoo Kim, and Qing Wang paper, slides, poster, code

Recent Posts