Dongxu Zhang · 张栋旭

Incoming Ph.D. student · Tsinghua University

Starting September 2027 · Advisor: Prof. Chang Qiao

I am a master's student in Software Engineering at Xi'an Jiaotong University, advised by Prof. Jihua Zhu. I will join Tsinghua University in September 2027 to pursue a Ph.D. under the supervision of Prof. Chang Qiao.

I received my B.Eng. in Computer Science and Technology from Taiyuan University of Technology, where I was advised by Prof. Juanjuan Zhao. I previously visited Shanghai AI Laboratory, working with Cheng Tan, and was a research intern with Ning Yang at the Institute of Automation, Chinese Academy of Sciences.

My research interests include efficient and reliable reasoning in large language models, multimodal learning, and AI for scientific discovery. My earlier work focused on 3D vision and point cloud registration.

News

Oct. 2026I will join Tsinghua University in September 2027 to pursue a Ph.D. under the supervision of Prof. Chang Qiao.
2026Our paper “Intern-Atlas: A Methodological Evolution Graph as Research Infrastructure for AI Scientists” has been accepted to NeurIPS 2026.
2026Our paper “VICT: Verifier-Instrumented Credit Tracing for Long-Horizon LLM Agent Reinforcement Learning” has been accepted to EMNLP 2026.
Jun. 2026Our paper “Not All Queries Need Deep Thought: CoFiCot for Adaptive Coarse-to-fine Stateful Refinement” has been accepted to UAI 2026.
Jun. 2026Our paper “Tri-Efficient Transfer Learning for Point Cloud Videos” has been accepted to ECCV 2026.
May 2026Our paper “Robust Regularized Policy Iteration under Transition Uncertainty” has been accepted to ICANN 2026.
May 2026Our paper “Intern-Atlas: A Methodological Evolution Graph as Research Infrastructure for AI Scientists” was featured on Hugging Face Daily Papers on May 1.

Earlier updates

Apr. 2026Our 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 (main conference, oral presentation).
Mar. 2026Our paper “PointRFT: Explicit Reinforcement Fine-tuning for Point Cloud Few-shot Learning” has been accepted to ICME 2026.
Mar. 2026Our paper “PersonalQ: Select, Quantize, and Serve Personalized Diffusion Models for Efficient Inference” has been accepted to ICME 2026.
Mar. 2026Our paper “PointCoT: A Multi-modal Benchmark for Explicit 3D Geometric Reasoning” is available on arXiv.
Jan. 2026Our paper “IGASA: Integrated Geometry-Aware and Skip-Attention Modules for Enhanced Point Cloud Registration” has been accepted by IEEE TCSVT.
Jan. 2026Our paper “CMHANet: A Cross-Modal Hybrid Attention Network for Point Cloud Registration” has been accepted by Neurocomputing.
Jan. 2026Our paper “Deep Local Field Consistency for Non-Rigid Point Cloud Registration” has been accepted to ICASSP 2026.

Publications

* Equal contribution · † Corresponding author

[NeurIPS 2026] Intern-Atlas: A Methodological Evolution Graph as Research Infrastructure for AI Scientists. Yujun Wu^*, Dongxu Zhang^*, Xinchen Li, et al.
[EMNLP 2026] VICT: Verifier-Instrumented Credit Tracing for Long-Horizon LLM Agent Reinforcement Learning. Pengcheng Li, Zhengyang Zhang, Dongxu Zhang, Sui Huang, Shaohua Ma.
[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^*, Jihua Zhu, Haozhe Cheng, Zhengqiao Li, Pengcheng Li, Chaowei Fang, Yonghao Dong, Lin Chen.
[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^*, Yiding Sun^*, Cheng Tan, Wenbiao Yan, Ning Yang, Jihua Zhu, Haijun Zhang.
[ICME 2026] PointRFT: Explicit Reinforcement Fine-tuning for Point Cloud Few-shot Learning. Yankai Wang, Yiding Sun, Qirui Wang, Pengbo Li, Chaoyi Lu, Dongxu Zhang.
[ICME 2026] PersonalQ: Select, Quantize, and Serve Personalized Diffusion Models for Efficient Inference. Qirui Wang, Qi Guo, Yiding Sun, Junkai Yang, Dongxu Zhang, Shanmin Pang, Qing Guo.
[arXiv 2026] PointCoT: A Multi-modal Benchmark for Explicit 3D Geometric Reasoning. Dongxu Zhang^*, Yiding Sun^*, Pengcheng Li^*, et al.
[Neurocomputing 2026] CMHANet: A Cross-Modal Hybrid Attention Network for Point Cloud Registration. Dongxu Zhang, Yingsen Wang, Yiding Sun, Haoran Xu, Peilin Fan, Jihua Zhu.
[IEEE TCSVT 2026] IGASA: Integrated Geometry-Aware and Skip-Attention Modules for Enhanced Point Cloud Registration. Dongxu Zhang, Jihua Zhu, Shiqi Li, Wenbiao Yan, Haoran Xu, Peilin Fan, Huimin Lu.
[ICASSP 2026] 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.

Open-source projects

NanoResearch
Jan. 2026 — Present

Core contributor

Contributed to an autonomous research workflow covering literature review, experiment planning, code generation, GPU and SLURM execution, and manuscript drafting. Worked on reusable research skills, project memory, workflow validation, and application demonstrations.
agentUniverse
Mar. 2025 — Present

Open-source contributor

Developed an image-to-text tool using OpenCV, EAST, and Tesseract, with Chinese and English recognition, image preprocessing, and fallback OCR. Added result cleaning, saving, and tests; the contribution was merged into the official repository in April 2025.