Jiazhou Zhou

About Me

Hi! I am Jiazhou Zhou (周佳舟), a PhD student (2024.09–2027.10) at AI Thrust, Information Hub of The Hong Kong University of Science and Technology (Guangzhou). I am fortunate to be advised by Prof. Lei Zhang and Prof. Yingcong Chen. I am also a research intern at International Digital Economy Academy (IDEA). I expect to graduate in October 2027 and am open to both industry and research positions.

Research Interests

Dedicated to building Next-Generation Self-Evolving Embodied Cognitive Systems grounded in physical reality. My research establishes a closed loop of Physical Interaction — World Modeling — Policy Self-Evolution to transcend static pre-training paradigms, focusing on three core thrusts:

Embodied Decision-Making & Policy Self-Evolution: Investigating active intervention and long-horizon multi-turn planning for embodied agents. Specifically, addressing long-term temporal credit assignment under sparse rewards via on-policy self-distillation and multimodal agentic reinforcement learning, enabling continuous self-improvement without dense human supervision.
Multimodal World Models & Embodied Counterfactual Reasoning: Developing physics-grounded world models that serve as internal neural physics engines for embodied agents. Leveraging active interrogation and self-reflection to facilitate latent counterfactual imagination, spatiotemporal causal prediction, and test-time planning scaling to provide predictive priors for complex decision-making.
Embodied Full-Modal Perception Foundations: Designing modality-agnostic representations across vision, touch, 3D geometric priors, and neuromorphic event streams in dynamic, unstructured physical environments, establishing a low-latency, robust sensory foundation for embodied physical grounding.

Publications (Selected)

Selected publications (12 papers, 10 published/accepted). Please refer to my Google Scholar for a full paper list.

Part I: Multimodal Agents, Agentic RL & On-Policy Distillation

Preprint

PIVOT: Pivot-Aware On Policy Self Distillation for Multi-Turn VLM Agents

Jiazhou Zhou, Hu Zhou, Yucheng Chen, Jinyuan Qu, Ying-Cong Chen, Lei Zhang

Under review, 2026.

Page PDF

VLM Agents Agentic RL On-Policy Distillation

Part II: Multimodal Reasoning & Latent Space Reasoning

Preprint

V-Reflection: Transforming MLLMs from Passive Observers to Active Interrogators

Jiazhou Zhou, Yucheng Chen, Haoran Li, Qing Jiang, Hu Zhou, Ying-Cong Chen, Lei Zhang

Under review, 2026.

Homepage PDF Code Hugging Face

MLLM Latent Reasoning Visual Grounding
EMNLP 2026

UR²-MLLM: Uncertainty-aware Revisit Reasoning in Multimodal Large Language Models for Radiology Report Generation

Yucheng Chen, Yang Yu, Jiazhou Zhou, Yufei Shi, Yongying Lan, Yichi Zhang, Liyi Li, Si Yong Yeo

The 2026 Conference on Empirical Methods in Natural Language Processing, EMNLP, 2026.

PDF

MLLM Uncertainty Medical Imaging
CVPR 2024 Highlight

ExACT: Language-guided Conceptual Reasoning and Uncertainty Estimation for Event-based Action Recognition and More

Jiazhou Zhou, Xu Zheng, Yuanhuiyi Lyu, Lin Wang

The IEEE/CVF Conference on Computer Vision and Pattern Recognition, CVPR (CCF-A), 2024. Highlight (2.8%)

PDF Page

Uncertainty Conceptual Reasoning Event Camera

Part III: Multimodal Alignment, Perception & Event Vision

AAAI 2025

T-Rex-Omni: Integrating Negative Visual Prompt in Generic Object Detection

Jiazhou Zhou, Qing Jiang, Kanghao Chen, Lutao Jiang, Yuanhuiyi Lyu, Ying-Cong Chen, Lei Zhang

The 40th Annual AAAI Conference on Artificial Intelligence, AAAI (CCF-A), 2025.

PDF T-Rex Series Product Page

Open world Object Detection Negative Visual Prompt
NeurIPS 2025

PASS: Path-selective State Space Model for Event-based Recognition

Jiazhou Zhou, Kanghao Chen, Lei Zhang, Lin Wang

The Thirty-Ninth Annual Conference on Neural Information Processing Systems, NeurIPS (CCF-A), 2025.

Page PDF Code

Event Camera State Space Model Recognition
CVPR 2024

UniBind: LLM-Augmented Unified and Balanced Representation Space to Bind Them All

Yuanhuiyi Lyu, Xu Zheng, Jiazhou Zhou, Lin Wang

The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) (CCF-A), 2024.

Page PDF Code

Multimodal Alignment Representation Learning
NeurIPS 2024

LaSe-E2V: Towards Language-guided Semantic-Aware Event-to-Video Reconstruction

Kanghao Chen, Haoran Li, Jiazhou Zhou, Zeyu Wang, Lin Wang

The Thirty-Eighth Annual Conference on Neural Information Processing Systems, NeurIPS (CCF-A), 2024.

PDF Page

Event-to-Video Language-guided
ECCV 2024

EventBind: Learning a Unified Representation to Bind Them All for Event-based Open-world Understanding

Jiazhou Zhou, Xu Zheng, Yuanhuiyi Lyu, Lin Wang

European Conference on Computer Vision, ECCV (CCF-B), 2024.

Page PDF Code

Event Camera Open-world CLIP
ECCV 2024

Centering the Value of Every Modality: Towards Efficient and Resilient Modality-agnostic Semantic Segmentation

Xu Zheng, Yuanhuiyi Lyu, Jiazhou Zhou, Lin Wang

European Conference on Computer Vision, ECCV (CCF-B), 2024.

PDF Page

Semantic Segmentation Multimodal

Professional & Academic Experience

[2025.08 - Present]: Research Intern, CVR Group, International Digital Economy Academy (IDEA)
[2023.06 - 2024.09]: Research Assistant, VLIS Lab, The Hong Kong University of Science and Technology
- Supervisor: Addison, WANG Lin

Services

Reviewer:
- Conferences: ICLR, CVPR, ECCV, ICCV, NeurIPS, AAAI, COLM, WACV, BMCV
- Journals: IEEE Transactions on Neural Networks and Learning Systems (TNNLS), IEEE Transactions on Intelligent Transportation Systems (TITS)
Teaching Assistant:
- UFUG2102, Matrix Algebra and Applications, HKUST(GZ), 2025.6-2025.7
- CMAA6102, Computational Media and Arts Program Seminar, HKUST(GZ), 2024.9-2025.5