Liyun Zhu

Liyun Zhu

Quantitative Researcher @ HiThink

Liyun Zhu

I am a Quantitative Researcher at HiThink Research, where I work on quantitative investment strategies and AI systems for trading. I received my master’s degree in Machine Learning and Computer Vision from the Australian National University. My research interests span LLM and agent systems for quantitative finance, multimodal large language models, and video understanding.

Research Interests

  • Agent systems for quantitative finance, including factor mining, event-driven analysis, trading signal generation, and portfolio optimization.
  • Multimodal large language models, video understanding, and anomaly detection.

Selected Publications

  1. Zhu, L., Wang, L., Raj, A., Gedeon, T., & Chen, C. (2024). Advancing Video Anomaly Detection: A Concise Review and a New Dataset. NeurIPS 2024 [Paper] [Website]

  2. Ding, D., Wang, L., Zhu, L., Gedeon, T., & Koniusz, P. (2024). LEGO: Learnable Expansion of Graph Operators for Multi-Modal Feature Fusion. ICLR 2025 [Paper]

  3. Zhang, C.*, Gan, Z.*, Zhu, L.*, Pang, Y., Zhang, Q., & Zhang, R. (2026). FinMTM: A Multi-Turn Multimodal Benchmark for Financial Reasoning and Agent Evaluation. arXiv 2026 [Paper] [Code]

  4. Zhu, L., Chen, Q., Shen, X., & Cun, X. (2025). VAU-R1: Advancing Video Anomaly Understanding via Reinforcement Fine-Tuning. arXiv 2025 [Paper] [Code]

* denotes equal contribution.

Service

  • Journal Reviewer: ACM Computing Surveys, IEEE TNNLS, Neurocomputing.

  • Conference Reviewer: NeurIPS 2026, AAAI 2026, NeurIPS 2025, ACM MM 2024, ICIP 2024.

Contact