Portrait
Zhuoyi (Jeremy) Peng
Ph.D. Candidate in Information Systems
HKUST Business School
Research Interests
Graph Machine Learning
Regulatory Machine Learning
About Me  — I study 1) LLM and Agent for Graph; 2) Machine Learning under Regulation

I am a Ph.D. candidate in Information Systems at the HKUST Business School. Before that, I received my M.Phil. in Computer Science and Engineering from HKUST and my B.Eng. in Software Engineering from Sun Yat-sen University.

My research develops trustworthy and knowledge-aware machine learning for real-world data and regulatory settings, along two main streams: (1) Relational Graph Learning — LLM agents that reason over graphs, graph–LLM co-teaching, and structured representation learning; and (2) Regulatory Machine Learning — machine unlearning, patent intelligence, and privacy-aware learning for compliance-sensitive applications such as finance and anti-fraud. My first-authored work has appeared at ICLR, NeurIPS, NAACL and IEEE Transactions on Big Data.

Education
  • The Hong Kong University of Science and Technology
    HKUST Business School
    Ph.D. in Information Systems
    Sep. 2023 - present
  • The Hong Kong University of Science and Technology
    Dept. of Computer Science and Engineering
    M.Phil. in Computer Science and Engineering
    Feb. 2019 - Nov. 2021
  • Sun Yat-sen University
    School of Software Engineering
    B.Eng. in Software Engineering
    Aug. 2014 - Jun. 2018
Industry Experience
  • WeBank
    Research Intern
    Jun. 2025 - Apr. 2026
  • HKUST Business School
    Research Assistant
    Apr. 2022 - Sep. 2023
  • Clustar
    Research Intern
    Feb. 2020 - Jan. 2021
  • Clustar
    Research Intern
    Apr. 2018 - Jan. 2019
  • Tencent
    Product Manager Intern
    Nov. 2017 - Feb. 2018
Honors & Awards
  • First-Class Academic Scholarship, Sun Yat-sen University
    2014 - 2018
News
2026
Beyond the Golden Teacher: Enhancing Graph Learning through LLM–GNN Co-teaching was accepted to NeurIPS 2026.
Sep 25
New preprints on LLM-for-graph reasoning are out, including GraphInfer-Bench, Agentic Graph Token Reasoning, and Beyond the Golden Teacher (LLM–GNN co-teaching).
Jun 15
2025
Started a research internship at WeBank, building large-scale graph systems for anti-fraud.
Jun 01
Adversarial Mixup Unlearning was accepted to ICLR 2025.
Jan 22
2024
Connecting the Dots: Inferring Patent Phrase Similarity with Retrieved Phrase Graphs was accepted to NAACL 2024 (Findings).
Mar 15
Selected Publications (view all )
PatentsNET: Graph-Based Pretraining for Patent Analytics

Zhitao Yin, Yi Yang, Zhuoyi Peng, Zhenghan Zhang

Working paper. Major revision at MIS Quarterly Major Revision

The first graph-pretraining framework for large-scale patent networks.

PatentsNET: Graph-Based Pretraining for Patent Analytics

Zhitao Yin, Yi Yang, Zhuoyi Peng, Zhenghan Zhang

Working paper. Major revision at MIS Quarterly Major Revision

The first graph-pretraining framework for large-scale patent networks.

Agentic Graph Token Reasoning

Zhuoyi Peng, Yi Yang

Preprint. arXiv preprint 2026

A state-of-the-art graph-native agent for graph analysis, featuring token-robust graph-LLM alignment, trajectory SFT, and preference optimization.

[arXiv]

Agentic Graph Token Reasoning

Zhuoyi Peng, Yi Yang

Preprint. arXiv preprint 2026

A state-of-the-art graph-native agent for graph analysis, featuring token-robust graph-LLM alignment, trajectory SFT, and preference optimization.

[arXiv]

GraphInfer-Bench: Benchmarking LLM's Inference Capability on Graphs

Zhuoyi Peng, Jingzhou Jiang, Hanlin Gu, Lixin Fan, Yi Yang

Preprint. arXiv preprint 2026

The first comprehensive benchmark for graph-LLM inference.

[arXiv]

GraphInfer-Bench: Benchmarking LLM's Inference Capability on Graphs

Zhuoyi Peng, Jingzhou Jiang, Hanlin Gu, Lixin Fan, Yi Yang

Preprint. arXiv preprint 2026

The first comprehensive benchmark for graph-LLM inference.

[arXiv]

Beyond the Golden Teacher: Enhancing Graph Learning through LLM-GNN Co-teaching

Zhuoyi Peng, Hanlin Gu, Lixin Fan, Yi Yang

Annual Conference on Neural Information Processing Systems (NeurIPS) 2026

Graph learning with sparse labels through co-teaching-based preference optimization.

[arXiv]

Beyond the Golden Teacher: Enhancing Graph Learning through LLM-GNN Co-teaching

Zhuoyi Peng, Hanlin Gu, Lixin Fan, Yi Yang

Annual Conference on Neural Information Processing Systems (NeurIPS) 2026

Graph learning with sparse labels through co-teaching-based preference optimization.

[arXiv]

Adversarial Mixup Unlearning

Zhuoyi Peng, Yixuan Tang, Yi Yang

International Conference on Learning Representations (ICLR) 2025

The first mixup-based method for machine unlearning.

[OpenReview]

Adversarial Mixup Unlearning

Zhuoyi Peng, Yixuan Tang, Yi Yang

International Conference on Learning Representations (ICLR) 2025

The first mixup-based method for machine unlearning.

[OpenReview]

Federated Meta Embedding Concept Stock Recommendation

Zhuoyi Peng, Yi Yang, Liu Yang, Kai Chen

IEEE Transactions on Big Data (TBD) 2024

A federated learning solution for financial analytics.

[IEEE Xplore]

Federated Meta Embedding Concept Stock Recommendation

Zhuoyi Peng, Yi Yang, Liu Yang, Kai Chen

IEEE Transactions on Big Data (TBD) 2024

A federated learning solution for financial analytics.

[IEEE Xplore]

Connecting the Dots: Inferring Patent Phrase Similarity with Retrieved Phrase Graphs

Zhuoyi Peng, Yi Yang

Findings of the North American Chapter of the Association for Computational Linguistics (NAACL Findings) 2024

One of the earliest applications of graph retrieval to domain-specific problems.

[ACL Anthology]

Connecting the Dots: Inferring Patent Phrase Similarity with Retrieved Phrase Graphs

Zhuoyi Peng, Yi Yang

Findings of the North American Chapter of the Association for Computational Linguistics (NAACL Findings) 2024

One of the earliest applications of graph retrieval to domain-specific problems.

[ACL Anthology]

All publications