昆士兰大学讲师蔡雨君博士学术讲座通知
报告题目：超越单一智能:多模态 AI助手中的协作力量
Beyond Single Intelligence: The Power of Collaboration in Multi-Modal AI Assistants
报 告 人：蔡雨君，昆士兰大学电气工程与计算机科学学院讲师（助理教授）
时间：2025年5月15日，周四上午9:30-12:00
地点：西五楼 404
报告人简介：
蔡雨君博士现为昆士兰大学（The University of Queensland）电气工程与计算机科学学院讲师（助理教授）。此前，她曾在美国 Meta 公司担任研究科学家，并于2021 年博士毕业于南洋理工大学。她的研究方向包括多模态感知与视觉-语言理解在顶级会议和期刊发蔡雨君表论文 30 余篇,总引用量超过 3800 次。她曾担任 NeurlPSICML、WWW、ICCV、ACM MM 和 WACV 等国际顶会的领域主席,以及 T-PAMIIJCV、T-IP、T-CSVT 等期刊的审稿人。
Yujun Cai is currently a Lecturer (Assistant Professor) at the School of Electrical Engineering and Computer Science, the University of Queensland (UQ). Before that she was a Research Scientist in Meta Inc USA. She obtained her Ph.D. from Nanyang Technological University in 2021. Her research focuses on multi-modal perception and vision-language understanding, with 30+ publications in top-tier venues and 3,800+ citations. She served as an Area Chair for conferences such as Neurips, ICML, WWW, ICCV, ACM MM, and WACV, and as a reviewer for journals such as T-PAMI, IJCV, T-IP, and T-CSVT.
报告摘要：
随着人工智能的不断发展，单一模型或模态在面对复杂的现实世界挑战时常常显得力不从心。本次报告将探讨如何通过模型之间以及态之间的协作，开启多态助手发展的新范式。通过整合不同专长的模型与多模态信息，我们可以增强推理能力、鲁棒性与适应性，从而突破单一模型的局限。然而，有效的协作也带来了模态对齐、计算效率与安全性等方面的挑战。报告将结合大模型协作与跨态交互的研究成果，展望多模态 AI助手朝着更灵活、更可扩展、更具鲁棒性的人类智能边进的未来。
As Artificial Intelligence continues to advance, single models or modalities sometimes face limitations in complex real-world challenges. This talk explores how collaboration across models and modalities creates a new paradigm for multi-modal assistant development. By integrating specialized models and multi-modal information, we can enhance reasoning capabilities, robustness, and adaptability while overcoming individual model limitations. However, effective collaboration introduces challenges in modal alignment, computational efficiency, and security. Drawing insights from large model cooperation and cross-modal interactions, we aim to advance multi-modal AI assistants towards the flexibility, scalability, and robust characteristic of human intelligence.
