中文

教育背景

我的研究兴趣位于多模态学习、基础模型、人本智能以及面向智能系统的软件工程的交叉方向。总体而言，我关注如何构建能够高效适配、稳健泛化、可信可控、与人自然协作，并能够在真实环境中被可靠构建、部署与维护的智能系统。

过去几年，我主要研究视觉语言模型的适配、参数高效微调与泛化，重点关注基础模型在新任务、新领域和分布变化条件下，如何以更低成本实现高效迁移与稳健泛化。在这一过程中，我尤其关注结合几何、优化与结构化建模等数学工具，探索更具原理性和可解释性的多模态学习方法。

近期，我也开始关注大模型的可信性与可控性，包括 Machine Unlearning 和 Watermarking & Fingerprinting 等问题，希望推动大模型在真实应用中的安全部署、内容溯源与可靠使用。

面向未来，我尤其对第一人称视觉、第一人称智能体以及第一人称多模态大模型感兴趣。我的长远目标是构建 Human-AI Co-Embodied Intelligence，使人工智能能够从第一人称视角理解人的行为、意图与环境交互，并在人机共处的真实世界中与人共同感知、共同理解、共同演化。

更广义地说，我的软件工程背景也促使我持续思考：如何将基础模型、多模态智能与具身智能的研究进展，进一步转化为可构建、可评测、可部署、可维护的真实智能系统。

多模态基础模型的高效适配与泛化：研究视觉语言模型与多模态基础模型的高效适配、参数高效微调与稳健泛化，包括 LoRA、Adapters、Prompt Learning 等参数高效方法。
大模型的可信与可控机制：研究 Machine Unlearning、Watermarking & Fingerprinting 等方向，提升大模型在真实应用中的可控性、可追踪性与可靠性。
第一人称多模态智能与人机共具身智能：研究第一人称视觉、第一人称智能体和第一人称多模态大模型，探索以人为中心的感知、推理与协作机制，推动 Human-AI Co-Embodied Intelligence。
面向多模态与具身智能系统的软件工程：研究如何将多模态基础模型与具身智能能力系统化地组织为可构建、可测试、可部署、可维护的真实系统，重点关注模块化架构、可靠性评测、人机闭环交互与生命周期管理。

	Gaussian Shannon: High-Precision Diffusion Model Watermarking Based on Communication Yi Zhang, Hongbo Huang, Liang-Jie Zhang The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR2026 ).CCF-A [论文][代码][项目网页 ]

	TLoRA: Task-aware Low Rank Adaptation of Large Language Models Weicheng Lin* *Yi Zhang(Co-first Author), Jiawei Dang, Liang-Jie Zhang The 64th Annual Meeting of the Association for Computational Linguistics (ACL2026** ).CCF-A [论文][代码][项目网页 ]

	Feature Projection Learning for Better Vision-Language Reasoning Yi Zhang , Weicheng Lin, Liang-Jie Zhang 2026 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP2026) CCF-B [论文]

	Training-Free Test-Time Adaptation with Brownian Distance Covariance in Vision-Language Models Yi Zhang , Chun-Wun Cheng, Angelica I Aviles-Rivero, Zhihai He, Liang-Jie Zhang 2026 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP2026) CCF-B [论文]

	Training-Free Dual Hyperbolic Adapters for Better Cross-Modal Reasoning Yi Zhang, Chun-Wun Cheng, Junyi He, Ke Yu, Yushun Tang, Carola-Bibiane Schönlieb, Angelica I Aviles-Rivero IEEE Transactions on Mutltimedia(TMM).CCF-A SCI中科院一区 [论文]

	Cross-Modal Few-Shot Learning with Second-Order Neural Ordinary Differential Equations Yi Zhang, Chun-Wun Cheng, Junyi He, Zhihai He, Carola-Bibiane Schönlieb, Yuyan Chen, Angelica I Aviles-Rivero AAAI Conference on Artificial Intelligence (AAAI2025). Oral Paper, 接收率=4.6\%, CCF-A [论文][Code][Project Page ]

	Training-Free Feature Reconstruction with Sparse Optimization for Vision-Language Models Yi Zhang, Ke Yu, Angelica I Aviles-Rivero, Jiyuan Jia, Yushun Tang, Zhihai He ACM International Conference on Multimedia (ACMMM2024) CCF-A [论文]

	Concept-Guided Prompt Learning for Generalization in Vision-Language Models Yi Zhang, Ce Zhang, Ke Yu, Yushun Tang, and Zhihai He The AAAI Conference on Artificial Intelligence(AAAI2024), CCF-A [论文]

	Conceptual Codebook Learning for Vision-Language Models Yi Zhang , Ke Yu, Siqi Wu and Zhihai He European Conference on Computer Vision (ECCV2024) CCF-B [论文]

	Test-time Distribution Learning Adapter for Cross-Modal Visual Reasoning Yi Zhang , Ce Zhang 2024 IEEE International Conference on Acoustics, Speech, and Signal Processing (ICASSP2024) CCF-B [论文]

	Cross-Model concept learning and reference for Vision-Language Models Yi Zhang , Ce Zhang, Yushun Tang, Zhihai He Neurocomputing. SCI 中科院二区 [论文]

	BDC-Adapter: Brownian Distance Covariance for Better Vision-Language Reasoning Yi Zhang , Ce Zhang, Zihan Liao, Yushun Tang, Zhihai He British Machine Vision Conference (BMVC2023). CCF-C [论文]

	Unsupervised Prototype Adapter for Vision-Language Models Yi Zhang , Ce Zhang, Xueting Hu, Zhihai He The 6th Chinese Conference on Pattern Recognition and Computer Visione (PRCV2023). CCF-C [论文]