I am currently a Master’s student at the Department of Software Engineering, Xi’an Jiaotong University (XJTU), advised by Prof. Jihua Zhu. I graduated from School of Computer Science and Technology, Taiyuan University of Technology with a bachelor’s degree. I am collaborating closely with Ning Yang at the Institute of Automation, Chinese Academy of Sciences (CASIA). I am currently a Visiting Student at the Shanghai Artificial Intelligence Laboratory (Shanghai AI Lab), advised by Cheng Tan, focusing on AI for Science and LLM Reasoning.

🚀I am now working on LLM Reasoning and Chain-of-Thought (CoT). Currently, I am actively seeking Ph.D. opportunities for the Fall 2027 intake. I am particularly interested in advancing LLM Reasoning and Multimodal Learning. If you are looking for a self-motivated student with a strong background in 3D-Vision and LLM, or if you are interested in academic collaboration, I would be honored to discuss further via email at zhangdongxu@stu.xjtu.edu.cn.

🔥 News

[Jun. 2026] 🎉 Excited to share that our paper “Not all queries need deep thought: Coficot for adaptive coarse-to-fine stateful refinement” has been accepted to UAI 2026! I am honored to be the first author.

[Jun. 2026] 🎉 Excited to share that our paper “Tri-Efficient Transfer Learning for Point Cloud Videos” has been accepted to ECCV 2026! I am honored to be co-first author (with Yiding Sun).

[May 2026] 🎉 Excited to share that our paper “Robust Regularized Policy Iteration under Transition Uncertainty” has been accepted to ICANN 2026! I serve as the corresponding author.

[Apr. 2026] 🎉 Excited to share that our paper “Chain-of-Thought Compression Should Not Be Blind: V-Skip for Efficient Multimodal Reasoning via Dual-Path Anchoring” has been accepted to ACL 2026 as a Main Conference Oral presentation! We propose V-Skip to address “Visual Amnesia” in MLLM reasoning compression. Paper | project page.

[Mar. 2026] 🎉 Excited to share that our two papers “PointRFT: Explicit Reinforcement Fine-tuning for Point Cloud Perception” and “PersonalQ: Select, Quantize, and Serve Personalized Diffusion Models for Efficient Inference” have been accepted by ICME 2026!

[Mar. 2026] 🔥 Excited to share our new preprint “PointCoT: A Multi-modal Benchmark for Explicit 3D Geometric Reasoning” on arXiv! We propose a novel Look–Think–Answer paradigm that empowers MLLMs with geometry-grounded Chain-of-Thought reasoning over 3D point clouds, significantly reducing geometric hallucinations. Check out the paper and project page!

[Jan. 2026] 🎉 Excited to share that our paper “IGASA: Integrated Geometry-Aware and Skip-Attention Modules for Enhanced Point Cloud Registration” has been accepted by IEEE TCSVT!

[Jan. 2026] 🎉 Excited to share that our paper “CMHANet: A Cross-Modal Hybrid Attention Network for Point Cloud Registration” has been accepted by Neurocomputing!

[Jan. 2026] 🎉 Excited to share that our paper “DEEP LOCAL FIELD CONSISTENCY FOR NON-RIGID POINT CLOUD REGISTRATION” has been accepted by ICASSP!

📝 Publications

UAI 2026 Not all queries need deep thought: Coficot for adaptive coarse-to-fine stateful refinement, Dongxu Zhang, Hongqiang Lin, Yiding Sun, Pengyu Wang, Qirui Wang, Ning Yang, Jihua Zhu.
ECCV 2026 Tri-Efficient Transfer Learning for Point Cloud Videos, Yiding Sun*, Dongxu Zhang*, et al.
ICANN 2026 Robust Regularized Policy Iteration under Transition Uncertainty, Hongqiang Lin, Zhenghui Fu, Weihao Tang, Pengfei Wang, Yiding Sun, Qixian Huang, Dongxu Zhang‡.
ACL 2026 (Main, Oral) Chain-of-Thought Compression Should Not Be Blind: V-Skip for Efficient Multimodal Reasoning via Dual-Path Anchoring, Dongxu Zhang, et al.
ICME 2026 PointRFT: Explicit Reinforcement Fine-tuning for Point Cloud Perception, Yiding Sun, Dongxu Zhang, et al.
ICME 2026 PersonalQ: Select, Quantize, and Serve Personalized Diffusion Models for Efficient Inference, Qirui Wang, Dongxu Zhang, et al.
ArXiv PointCoT: A Multi-modal Benchmark for Explicit 3D Geometric Reasoning, Dongxu Zhang*, Yiding Sun*, Pengcheng Li*, et al.
Neurocomputing CMHANet: A Cross-Modal Hybrid Attention Network for Point Cloud Registration, Dongxu Zhang, et al.
IEEE TCSVT IGASA: Integrated Geometry-Aware and Skip-Attention Modules for Enhanced Point Cloud Registration, Dongxu Zhang, Jihua Zhu, et al.
ICASSP Deep Local Field Consistency for Non-Rigid Point Cloud Registration, Haoran Xu, Dongxu Zhang, et al.
IEEE TIA Enhancing power grid resilience with blockchain-enabled vehicle-to-vehicle energy trading in renewable energy integration, Yingsen Wang, Dongxu Zhang, et al.

🎖 Honors and Awards

2024.12 National first prize in Tecorigin Operator Development Task Challenge.
2024.06 Outstanding graduate in TYUT.
2023.12 First-class scholarship (Top 5%)
2022.12 National Second prize in C++ group of Computer Ability Challenge.
2021.05 Truth-Seeking Star (10 students in TYUT each year)

📖 Educations

2024.09 - 2027.06 (now), Master, Software engineer, Xi’an Jiaotong University, Xi’an. Mentor: Jihua Zhu
2020.09 - 2024.06, Undergraduate, Computer science and technology, Taiyuan University of Technology, Taiyuan, Mentor: Juanjuan Zhao

💬 Invited Talks

2025.02, Taichu 2024 excellent developer oral report, Beijing. Info
2024.12, First prize defense in Tecorigin Operator Development Task Challenge, Wuhan. Info
2021.05, Outstanding student representative gave a speech, Taiyuan. Info

💻 Internships and Research