Welcome to Xingyue Huang’s Personal Website
Welcome! 👋
I’m Xingyue Huang (黄星越), a DPhil student in Computer Science at the University of Oxford, supervised by Prof. Michael M. Bronstein and Dr. İsmail İlkan Ceylan. My work focuses on language models, AI agents, and foundation models, with an emphasis on structured reasoning and generalization.
I work on language model pre-training, long-context modeling, and efficient attention, as well as agent systems and architectures for tool-using, reasoning-oriented AI. My research background is in graph machine learning, knowledge graphs, and relational learning — studying how models reason over structured data and generalize to new entities, relations, and tasks.
My work has been published at ACL, NeurIPS, ICML, and ICLR. I’m especially interested in building LLM-based systems that reason reliably and use context effectively, and I’m open to connecting with people working on language models, agents, foundation models, and applied AI.
Research interests: Large Language Models, AI Agents, Foundation Models, Long-Context Modeling, Structured Reasoning, Retrieval-Augmented Generation, Knowledge Graphs, Graph Neural Networks, Relational Learning.
📰 News
Selected as Silver Reviewer for ICML 2026
Published:
Honored to be recognized as an ICML 2026 Silver Reviewer.
RelAgent Released on arXiv
Published:
Our paper “RelAgent: LLM Agents as Data Scientists for Relational Learning” is now available on arXiv.
Workshop Proposal Accepted at ICML 2026
Published:
Our workshop proposal, “Graph Foundation Models: A New Era for Graph Machine Learning,” has been accepted for ICML 2026!
TDA Accepted to ACL 2026
Published:
Our paper “Threshold Differential Attention for Sink-Free, Ultra-Sparse, and Non-Dispersive Language Modeling” has been accepted to ACL 2026.
HTP Accepted to ACL 2026 (Oral)
Published:
Our paper “Hierarchical Token Prepending: Enhancing Information Flow in Decoder-based LLM Embeddings” has been accepted to ACL 2026 (Oral).
Selected as Top Reviewer for NeurIPS 2025
Published:
Honored to be recognized as a Top Reviewer at NeurIPS 2025.
Organizing the SEA Workshop @ NeurIPS 2025
Published:
Co-organizing the Scaling Environments for Agents (SEA) Workshop at NeurIPS 2025.
Research Intern at Snap Inc. (User Modelling & Personalization)
Published:
Research intern at Snap Inc., User Modelling and Personalization teams.
Research Internships at Eigent-AI & CAMEL-AI
Published:
Worked on multi-agent systems and data generation for verifiable RL.
How Expressive are Knowledge Graph Foundation Models? Permalink
Published:
Our paper ‘How Expressive are Knowledge Graph Foundation Models?’ is now available on arXiv. Check it out for a detailed analysis of the expressive power of KGFMs.
Line Graph Transformation Paper Accepted to NeurIPS 2024 Workshop
Published:
Our paper “Theoretical Insights into Line Graph Transformation on Graph Learning” was accepted to the NeurIPS 2024 Workshop of Symmetry and Geometry in Neural Representations.
AnyCQ Released on arXiv
Published:
Our paper “One Model, Any Conjunctive Query: Graph Neural Networks for Answering Complex Queries over Knowledge Graphs” is now available on arXiv.
Cooperative Graph Neural Networks Accepted to ICML 2024
Published:
Our paper “Cooperative Graph Neural Networks” was accepted to ICML 2024.
Genetic Programming Paper Accepted to IEEE Transactions on Cybernetics
Published:
Our paper “A Novel Multiobjective Genetic Programming Approach to High-Dimensional Data Classification” was accepted to IEEE Transactions on Cybernetics.
Relational Hypergraphs Paper Released on arXiv
Published:
Our paper “Link Prediction with Relational Hypergraphs” is now available on arXiv.
A Theory of Link Prediction Accepted to NeurIPS 2023
Published:
Our paper “A Theory of Link Prediction via Relational Weisfeiler-Leman on Knowledge Graphs” was accepted to NeurIPS 2023.
🛠️ Service
- Reviewer: NeurIPS 2025 Top Reviewer, ICML 2026 Silver Reviewer, ICLR 2026
- Workshop Organizer: NeurIPS 2025 SEA, ICML 2026 GFM