Wenhui Zhu (朱文辉)

I am a Ph.D. Candidate at the School of Computing and Augmented Intelligence (SCAI), Arizona State University, advised by Prof. Yalin Wang.

My research drives the full lifecycle of Generative AI, specializing in Efficient LLMs/VLMs/MLLMs, web-scale multimodal data curation, foundation model scaling, post-training alignment, and agentic reinforcement learning.

I design autonomous, long-horizon, Self-Evolving LLM/MLLMs agents capable of complex reasoning by stabilizing advanced post-training at scale and developing efficient training-free inference-time optimizations.

My findings are regularly published at top-tier venues including ICLR, EMNLP, ICML, ACL, CVPR, ECCV, ICCV, and COLM.

Email: wzhu59@asu.edu

Google Scholar  /  GitHub  /  LinkedIn  /  X  /  CV

profile photo

📍 Tempe, Arizona, USA

News

  • 🏆 May 2026: Honored to receive the Outstanding Graduate Student Award at ASU SCAI. Only one student is selected per department per year.

  • 🔥 May 2026: Our paper S-SPPO: Semantic-Calibrated Self-Play Preference Optimization was accepted by ICML 2026.

  • 💡 May 2026: Our papers DRA-GRPO and AHA were accepted by ACL 2026.

  • 🚀 May 2026: Released new work on agentic RL, coding-agent context pruning, and financial trading agents.

  • 🎯 Apr. 2026: Explored training-free reward-guided decoding via Sequential Monte Carlo and vulnerabilities in agentic LLMs.

  • 🌟 Feb. 2026: Our work on large language diffusion models, LLaDA-MedV, was accepted by CVPR 2025.

  • 🥇 Oct. 2025: Our team won 1st Place in the MICCAI MuCaRD 2025 Challenge.

  • 🌟 Feb. 2025: Our work on Multi-modal Representation Learning, Multimodal Variational Autoencoder: A Barycentric View, was select by AAAI 2025 Oral.

  • 👏 2024: DGR-MIL was accepted by ECCV 2024, and TimeMIL was accepted by ICML 2024.

Selected Publications (Full Publications)

🔥 Post-Training & Alignment

S-SPPO: Semantic-Calibrated Self-Play Preference Optimization [ICML 2026]
Peijie Qiu*, Wenhui Zhu*, Xiwen Chen, Jingjing Wang, Zhipeng Wang, Huayu Li, ZhengXiao He, Xuanzhao Dong, et al.
International Conference on Machine Learning (ICML), 2026. * Equal contribution.
DRA-GRPO: Exploring Diversity-Aware Reward Adjustment for R1-Zero-Like Training of Large Language Models [ACL 2026]
Xiwen Chen*, Wenhui Zhu*, Peijie Qiu, Xuanzhao Dong, Haoran Wang, Hong Wu, Huayu Li, Aristeidis Sotiras, Yalin Wang, et al.
Annual Meeting of the Association for Computational Linguistics (ACL), 2026. [Paper] * Equal contribution.
Sampling for Quality: Training-Free Reward-Guided LLM Decoding via Sequential Monte Carlo [COLM 2026]
J. Markovic-Voronov*, Wenhui Zhu*, Bo Long, Zhen Wang, Shixiang Shane Gu, Kayhan Behdin, et al.
Conference on Language Modeling (COLM), 2026. [Paper] * Equal contribution.
Prompt-OT: An Optimal Transport Regularization Paradigm for Knowledge Preservation in Vision-Language Model Adaptation [WACV 2026]
Xiwen Chen*, Wenhui Zhu*, Peijie Qiu, Haoran Wang, Huayu Li, Hong Wu, Xuanzhao Dong, Aristeidis Sotiras, Yalin Wang, et al.
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2026. * Equal contribution.
AHA: Aligning Large Audio-Language Models for Reasoning Hallucinations via Counterfactual Hard Negatives [ACL 2026]
Yanxi Chen, Wenhui Zhu*, Xiwen Chen, Zhipeng Wang, Xin Li, Peijie Qiu, Hao Wang, Xuanzhao Dong*, Yujian Xiong, Anderson Schneider, Yuriy Nevmyvaka, Yalin Wang.
Annual Meeting of the Association for Computational Linguistics (ACL), 2026. * Equal contribution.

🤖 Agentic Reinforcement Learning & Systems

AriadneMem: Threading the Maze of Lifelong Memory for LLM Agents
Wenhui Zhu, Xiwen Chen, Zhipeng Wang, Jingjing Wang, Xuanzhao Dong, M. Huang, R. Cai, Huaijin Sang, et al.
arXiv preprint, 2026. [Paper]
Your Agent is More Brittle Than You Think: Uncovering Indirect Injection Vulnerabilities in Agentic LLMs
Wenhui Zhu, Xuanzhao Dong, Xiwen Chen, R. Cai, Peijie Qiu, Zhipeng Wang, O. Frunza, S. Tang, J. Gu, et al.
arXiv preprint, 2026. [Paper]
Mags-RL: Wearing Multimodal LLMs a Magnifying Glass via Agentic Reinforcement Learning for Complex Scene Reasoning
Xuanzhao Dong*, Wenhui Zhu*, Peijie Qiu, Xiwen Chen, Xiaobing Yu, Xin Li, Zhipeng Wang, Shao Tang, Gen Li, Yujian Xiong, Hao Wang, Yanxi Chen, Prayag Tiwari, Yalin Wang.
arXiv preprint, 2026. * Equal contribution.
Context Pruning for Coding Agents via Multi-Rubric Latent Reasoning
Jingjing Wang*, Xiwen Chen*, Wenhui Zhu*, Huayu Li, Zhipeng Wang, ZhengXiao He, F. Cai, A. S. Carreon-Rascon, Xuanzhao Dong, et al.
arXiv preprint, 2026. [Paper] * Equal contribution.
EZBlender: Efficient 3D Editing with Plan-and-ReAct Agent [WACV 2026 Oral]
Haoran Wang*, Wenhui Zhu*, S. Tang, Zhipeng Wang, Xuanzhao Dong, X. Li, Xiwen Chen, A. Bastola, et al.
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2026. * Equal contribution.

⚡ Efficient MLLMs/LLMs

CE-GAD: Curriculum-Efficient On-Policy Adversarial Distillation of Large Language Models
Wenhui Zhu, Hejian Sang, Xiwen Chen, Shayan Mohajer Hamidi, Han Yu, Zhipeng Wang, Jingjing Wang, Yuanda Xu, Xuanzhao Dong, Zhengze Zhou, Ran He, Jelena Markovic-Voronov, Kayhan Behdin, Sayan Ghosh, Alborz Geramifard.
Submission, 2026.
SODA: Semi On-Policy Black-Box Distillation for Large Language Models
Xiwen Chen*, Jingjing Wang*, Wenhui Zhu*, Peijie Qiu, Xuanzhao Dong, Hejian Sang, Zhipeng Wang, Alborz Geramifard, Feng Luo.
arXiv preprint, 2026. * Equal contribution.
OTPrune: Distribution-Aligned Visual Token Pruning via Optimal Transport [CVPR 2026]
Xiwen Chen*, Wenhui Zhu*, G. Li, Xuanzhao Dong, Yujian Xiong, Haoran Wang, Peijie Qiu, Q. Song, Zhipeng Wang, et al.
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2026. [Paper] * Equal contribution.
EVTP-IVS: Effective Visual Token Pruning for Unifying Instruction Visual Segmentation in Multi-Modal Large Language Models [WACV 2026]
Wenhui Zhu, Xiwen Chen, Zhipeng Wang, S. Tang, S. Ghosh, Xuanzhao Dong, R. Koner, Yalin Wang.
IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), 2026. [Paper]

MLLMs/Foundation Model and Large-Scale Data Curation Scaling

Ophth-500K: Curating Unstructured Video Streams for Scaling Multimodal Ophthalmology Foundation Models
Hao Wang, Xin Li, Wenhui Zhu*, Xuanzhao Dong, Xiwen Chen, Yujian Xiong, Langechuan Liu, Abolfazl Razi, Oana Dumitrascu, Yalin Wang.
arXiv preprint, 2026. * Equal contribution.
OphIn-500K: Curating Web-Scale Visual Instructions for Scaling Ophthalmic Multimodal Large Language Models
Xuanzhao Dong, Wenhui Zhu*, Xiwen Chen, Hao Wang, Xin Li, Yujian Xiong, Jiajun Cheng, Jingjing Wang, Xiaobing Yu, Haiyu Wu, Shao Tang, Zhipeng Wang, Langechuan Liu, Shan Lin, Oana Dumitrascu, Yalin Wang.
Submission, 2026. * Equal contribution.
LLaDA-MedV: Exploring Large Language Diffusion Models for Biomedical Image Understanding [CVPR 2025]
Xuanzhao Dong, Wenhui Zhu*, Xiwen Chen, Zhipeng Wang, Peijie Qiu, S. Tang, X. Li, Yalin Wang.
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025. [Paper] * Equal contribution.

Other Fields

Cracking Instance Jigsaw Puzzles: An Alternative to Multiple Instance Learning for Whole Slide Image Analysis [ICCV 2026]
Xiwen Chen, Peijie Qiu, Wenhui Zhu*, Hao Wang, Huayu Li, Xuanzhao Dong, Xiaotong Sun, Xiaobing Yu, Yalin Wang, Abolfazl Razi, Aristeidis Sotiras.
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). * Equal contribution.
FIC-TSC: Learning Time Series Classification with Fisher Information Constraint [ICML 2025]
Xiwen Chen, Wenhui Zhu*, P. Qiu, H. Wang, H. Li, Z. Li, Y. Wang, A. Sotiras, A. Razi.
International Conference on Machine Learning (ICML), 2025. [Paper] * Equal contribution.
How Effective Can Dropout Be in Multiple Instance Learning? [ICML 2025]
Wenhui Zhu, P. Qiu, Xiwen Chen, Z. Yang, A. Sotiras, A. Razi, Yalin Wang.
International Conference on Machine Learning (ICML), 2025.
DGR-MIL: Exploring Diverse Global Representation in Multiple Instance Learning for Whole Slide Image Classification [ECCV 2024]
Wenhui Zhu, Xiwen Chen, P. Qiu, A. Sotiras, A. Razi, Yalin Wang.
European Conference on Computer Vision (ECCV), 2024.
TimeMIL: Advancing Multivariate Time Series Classification via a Time-Aware Multiple Instance Learning [ICML 2024]
Xiwen Chen, P. Qiu, Wenhui Zhu*, H. Li, H. Wang, A. Sotiras, Yalin Wang, A. Razi.
International Conference on Machine Learning (ICML), 2024. * Equal contribution.

Invited Talks

  • Jan. 2026: Oral Presentation, WACV 2026.

  • Oct. 2025: Award Talk and Challenge Presentation, MICCAI MuCaRD Challenge, MICCAI 2025.

  • Oct. 2024: Oral Presentation, AAAI 2025.

Selected Honors & Awards

  • Outstanding Graduate Student Award, School of Computing and Augmented Intelligence (SCAI), ASU, 2026. Sole recipient per department.

  • 1st Place Winner, MICCAI Multi-Sequence Cardiac MRI Segmentation (MuCaRD) Challenge, Final Task, 2025.

  • Outstanding Contribution Award, MICCAI Ultra-Widefield Fundus Fluorescein Angiography for Diabetic Retinopathy (UWF4DR) Challenge, Tasks 1 and 3, 2024.

  • Two 3rd Place Awards, MICCAI UWF4DR Challenge, Tasks 1 and 3, 2024. Out of 22 teams.

  • Two 3rd Place Awards, MICCAI Multi-Modal Classification of Myopic Maculopathy (MMAC) Challenge, Tasks 1 and 3, 2023. Out of 59 teams.

  • SCAI Doctoral Fellowship, Arizona State University, Spring 2024.

  • ASU Graduate Research Fellowship / Scholarship.

  • Travel Awards, WACV Travel Award 2026 and Travel Awards by Arizona State University.

Academic Services

  • Conference Reviewer: ACL, EMNLP, NeurIPS, ICML, ICLR, CVPR, ICCV, ECCV, COLM, MICCAI, AAAI, and WACV.

  • Journal Reviewer: JAMA Ophthalmology, IEEE TMI, IEEE TPAMI, IEEE TAI, Medical Image Analysis (MedIA), and TMLR.

Teaching

  • Teaching Assistant, Ira A. Fulton Schools of Engineering, School of Computing and Augmented Intelligence, Arizona State University.

Flag Counter
© Wenhui Zhu | Last updated: May 2026