Xinyue XU

I am currently a PhD Candidate at The Hong Kong University of Science and Technology, advised by Prof. Xiaomeng Li and co-advised by Prof. Hao Wang at Rutgers University. Prior of this, I completed my Bachelor degree (with Honours) at The Australian National University, where I had the privilege of being advised by Senior Prof. Amanda Barnard and supervised by Dr. Amanda Parker. My research journey includes experiences as a Research Assistant at MIT's Computational Connectomics Group lead by Prof. Nir Shavit, where I had the opportunity to collaborate with Prof. Hao Wang and Prof. Lu Mi. Furthermore, I had served as a Research Assistant under the supervision of Dr. Gang Guo at NUS.

Email  /  CV  /  Google Scholar  /  LinkedIn

profile photo
Research

My research interest is to apply Machine Learning to interdisciplinary fields.
In particluar, I am interested in:

  • Interpretable ML (Concept-Based Explanations, Rule-Based Models, etc.)
  • AI4Science (medical image analysis, environmental science, material science etc.)
  • Multimodal LLM

Publications

"*" indicates equal contribution, "_" indicates equal advising.

Energy-Based Concept Bottleneck Models: Unifying Prediction, Concept Intervention, and Probabilistic Interpretations
Xinyue Xu, Yi Qin, Lu Mi, Hao Wang, Xiaomeng Li
ICLR, 2024
Paper/Code

Interpretability

We introduce Energy-Based Concept Bottleneck Models (ECBM) as a unified framework for concept-based prediction, concept correction, and fine-grained interpretations based on conditional probabilities.

Machine Learning Models for Evaluating Biological Reactivity within Molecular Fingerprints of Dissolved Organic Matter over time
Chen Zhao, Kai Wang, Qianji Jiao, Xinyue Xu, Yuanbi Yi, Penghui Li, Julian Merder, Ding He
Geophysical Research Letters, 2024
Paper

AI4Science Interpretability

Machine learning models were built to correlate the molecular composition and biological reactivity at the world's largest reservoir. Shorter incubations result in a broader range of molecules disappearing, with a greater contribution of stochasticity. Tuning the machine learning model contributes to yield additional interpretability beyond its well-recognized predictive power.

Dynamic Data Augmentation via MCTS for Prostate MRI Segmentation
Xinyue Xu*, Yuhan Hsi*, Haonan Wang, Xiaomeng Li
ICONIP, 2023
Paper/Code

AI4Science

Medical image data are often limited due to the expensive acquisition and annotation process. Hence, training a deep-learning model with only raw data can easily lead to overfitting. To this end, we present a novel method, called Dynamic Data Augmentation (DDAug), which is efficient and has negligible computation cost.

Exploring the Complexities of Dissolved Organic Matter Photochemistry from the Molecular Level by Using Machine Learning Approaches
Chen Zhao*, Xinyue Xu*, Hongmei Chen, Fengwen Wang, Penghui Li, Chen He, Quan Shi, Yuanbi Yi, Xiaomeng Li, Siliang Li, Ding He
Environmental Science & Technology (ES&T), 2023
The 6th Xiamen Symposium on Marine Environmental Sciences, 2023
Paper

AI4Science Interpretability

Photochemical reactions are essential components altering dissolved organic matter (DOM) chemistry. We first used machine learning approaches to compatibly integrate existing irradiation experiments and provide novel insights into the estuarine DOM transformation processes.

Subgraph frequency distribution estimation using graph neural network
Zhongren Chen*, Xinyue Xu*, Shengyi Jiang, Hao Wang, Lu Mi
KDD Deep Learning on Graphs, 2022
arXiv

AI4Science

Small subgraphs (graphlets) are important features to describe fundamental units of a large network. Unfortunately due to the inherent complexity of this task, most of the existing methods are computationally intensive and inefficient. In this work, we propose GNNS, a novel representational learning framework that utilizes graph neural networks to sample subgraphs efficiently for estimating their frequency distribution.

Classification Models for Medical Data with Interpretative Rules
Xinyue Xu, Xiang Ding, Zhenyue Qin, Yang Liu
ICONIP, 2021
Paper

Interpretability AI4Science

We use a variety of classification models to recognize the positive cases of SARS. We conduct evaluation with two types of SARS datasets, numerical and categorical types. For the sake of more clear interpretability, we also generate explanatory rules for the models.

Honours, Membership and Sevices
  • Hong Kong PhD Fellowship, 2022 - 2026
  • Conference Travel Allowance of HKPFS, 2023 - 2024
  • HKUST RedBird PhD Award, 2022 - 2024
  • IEEE Member, since 2021
  • Australian Computer Society Member, since 2021
  • Asia Pacific Neural Network Society Member, since 2021
  • ICLR Reviewer, 2025
  • ICONIP PC Member, 2023 - 2024
  • Teaching Assistant for ELEC 1200 A System View of Communications: from Signals to Packets, 2023

The source code is from Jon Barron's public academic website.