Tianhao Fu

I'm currently a quantitative researcher at a global hedge fund in Singapore. Prior to this, I graduated from Peking University with master degree. I also spent wonderful time at Microsoft Research Lab Asia, Sea AI Lab, Alibaba DAMO AI Academy, Baidu AI Research, Tencent Youtu AI Research, Ruitian Capital, and Nanyang Technological University/Tsinghua University.

Email / Scholar / Github

News

📈 Started new journey: Finding Alphas in US/CN/EU/JP region equity and crypto market, with some exciting results.

🥇 Won WorldQuant competition gold medal prize.

🥇 Won Microsoft Stars-of-Tomorrow prize.

🏁 Open Sourced Tiny3D: the world first Data-Centric 3D AI service production system.

🥇 Ranked Global 1st/1973 at CLUE LLM Benchmark.

🏁 Invited as a Program Committee for AAAI.

Selected Research

I'm interested in Large Language Model, Natural Language Processing and Computer Vision.

* means Equal Contribution.

	Large Language Model as an Excellent Reinforcement Learning Researcher in Financial Single-Agent and Multi-Agent Scenarios Tianhao Fu, Xinxin Xu, Weichen Xu, Ruilong Ren, Bowen Deng, Xinyu Zhao, Jian Cao, Xixin Cao Preprint, 2025 paper Use a self-evolutional LLM to automatically design and improve Reinforcement Learning Algorithm.
	Lifting the Veil of Non-Stationarity in Financial Market Tianhao Fu, Xinxin Xu, Xuanmeng Zhang, Weichen Xu, Ruilong Ren, Bowen Deng, Xinyu Zhao, Jian Cao, Xixin Cao Preprint, 2025 paper Use Market-State Jump Diffusion Process to capture financial market Non-Stationarity.
	Two Heads are Better than One: Distilling Large Language Model Features Into Small Models with Feature Decomposition and Mixture Tianhao Fu, Xinxin Xu, Weichen Xu, Ruilong Ren, Bowen Deng, Xinyu Zhao, Jian Cao, Xixin Cao Preprint, 2025 paper Distill decomposed LLM teacher feature into small student model.
	Mutual information-driven self-supervised point cloud pre-training Weichen Xu, Tianhao Fu, Jian Cao, Xinyu Zhao, Xinxin Xu, Xixin Cao, Xing Zhang Knowledge-Based Systems, 2025 paper Use mutual information to pre-train deep learning model in autonomous driving area.
	Empirical Research On Quantization For 3D Multi-Modal Vit Models Zicong Hu, Jian Cao, Weichen Xu, Ruilong Ren, Tianhao Fu, Xinxin Xu, Xing Zhang IEEE International Conference on Image Processing (ICIP), 2024 paper Bag of tricks for compressing and speeding up the deep learning model in autonomous driving area.
	DCLNet: Data Closed-Loop Network for Laryngoscopy Image Annotation and Classification Xinxin Xu, Xinyu Zhao, Jian Cao, Weichen Xu, Tianhao Fu, Ruilong Ren, Zicong Hu, Xing Zhang IEEE Biomedical Circuits and Systems Conference (BioCAS), 2024 paper Use LLM to do automatically data labeling to train a deep learning model in medical area.
	Boosting 3D Visual Grounding by Object-Centric Referring Network Ruilong Ren, Jian Cao, Weichen Xu, Tianhao Fu, Yilei Dong, Xinxin Xu, Zicong Hu, Xing Zhang IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2024 paper Use object referring information to do 3D Visual Grounding.
	J-MAE: Jigsaw Meets Masked Autoencoders in X-Ray Security Inspection Weichen Xu, Jian Cao, Tianhao Fu, Awen Bai, Ruilong Ren, Zicong Hu, Xixin Cao, Xing Zhang IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024 paper Through shuffling image patches to make deep learning model learn cross-area information.
	SweepMM: A High-Quality Multimodal Dataset for Sweeping Robots in Home Scenarios for Vision-Language Model Weichen Xu, Xinxin Xu, Tianhao Fu, Jian Cao, Xiaoyang Xu, Yuetian Huang, Xixin Cao, Xing Zhang IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2024 paper A dataset that could used for various sweeper robot task.
	Overcoming the seesaw in monocular 3d object detection via language knowledge transferring Weichen Xu, Tianhao Fu IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2023 paper Distill Large Language Model knowledge into 3d object detection model.
	Boosting dense long-tailed object detection from data-centric view Weichen Xu, Jian Cao, Tianhao Fu, Hongyi Yao, Yuan Wang Proceedings of the Asian Conference on Computer Vision, 2022 paper A new sampling method to train object detection model.
	Lifting the Veil of Frequency in Joint Segmentation and Depth Estimation Tianhao Fu, Yingying Li, Xiaoqing Ye, Xiao Tan, Hao Sun, Fumin Shen, Errui Ding Proceedings of the 29th ACM International Conference on Multimedia(MM), 2021 paper Decompose multi-task feature in the frequency domain.

Professional Services

Program Committee: AAAI 2022, AAAI 2023, AAAI2025
Reviewer: MM 2022, ICASSP 2023, ACCV 2023, IROS 2024, ICASSP 2024, ICIP 2024, ACCV 2024, IJCAI 2024, KDD 2024, MM 2024

Founded Startups

Fulcrum.ai: 2025-

TinyData AI(TinyData-ML): 2022-2023

Template from Jon Barron.