Research Interests
-
My research targets AI for Science & Bioinformatics (drug discovery), analyzing and interpreting biomedical data (molecules).
-
I focus on the information-theoretic foundations of machine learning: Invariant and Multi-view Representation Learning, Causal Inference, Information Bottleneck, Explainable AI, Pattern Recognition, and Fairness Learning.
-
I am interested in Graph Neural Networks, Transformers, Knowledge Graphs, and LLMs for knowledge discovery.
Selected Publications (Top-tier conferences/journals)
-
Van Thuy Hoang, O-Joun Lee.
Pre-training Graph Neural Networks on Molecules by using Subgraph-conditioned Graph Information Bottleneck.
In AAAI Conference on Artificial Intelligence 2025 (AAAI 2025). 📄PDF
Ranking A*
-
Van Thuy Hoang, Hyeon-Ju Jeon, O-Joun Lee.
Mitigating Degree Bias in Graph Representation Learning with Learnable Structural Augmentation and Structural Self-attention.
In IEEE Transactions on Network Science and Engineering 2025 (IEEE TNSE 2025). 📄PDF
Top 1.8% JCR
-
Van Thuy Hoang, O-Joun Lee.
Transitivity-Preserving Graph Representation Learning for Bridging Local Connectivity and Role-based Similarity.
In AAAI Conference on Artificial Intelligence 2024 (AAAI 2024). 📄PDF
Ranking A*
Research Philosophy
I view Wisdom —not simply Knowledge— as relational, arising from inter-dependent relationships rather than static/isolated entities.
Grounded in fundamental theories from both Buddhist insight and Christian thought, all existence flows from and returns to a unifying source.
These foundations align deeply with my theoretical foundations: Information Bottleneck and Causal Inference.
There is no central self —only the Original Mind, the Buddha-nature itself or Ground of Soul— emerging through a series of interdependent arisings.
The nature of reality thus unfolds as an invariant causal network.
Academic Services
-
Conference Reviewer: AAAI-26, ICLR-26
-
Journal Reviewer:
npj Computational Materials,
Pattern Recognition,
Journal of Big Data,
Neurocomputing,
Knowledge-Based Systems,
Expert Systems with Applications,
Information Sciences