Title: Multimodal LLM Alignment: Dimensions, Challenges, and Methods

Speaker: Lina Yao, University of New South Wales, Australia

Biography
Lina Yao is currently a Professor at UNSW. Her research spans machine learning, data mining, information retrieval, recommender systems, and natural language processing, with a particular focus on developing robust ML systems and intelligent AI agents. She is deeply interested in explainable and personalized AI, as well as human-AI cooperation, aiming to create steerable, user-centric, and impactful systems.

In this talk, she will highlight the key dimensions of multimodal LLM alignment, outline the main challenges in this area, and share some of our recent work. She will also point to open opportunities for further advancing multimodal alignment.

Title: Agentic FM Hands-on Tutorial

Speaker: Peng Yan, University of Technology Sydney, Australia

Biography
Dr. Peng Yan is a research fellow at the Australian Artificial Intelligence Institute (AAII), University of Technology Sydney (UTS). His research focuses on federated learning, model personalization, and model interpretability. Prior to joining UTS, he gained extensive industry experience at leading global consultancy firms, where he developed AI solutions in areas including information indexing, fraud detection, and recommendation algorithms.

The session will provide a brief overview of the evolution from Foundation Models to Agentic Foundation Models, explaining core concepts, motivations, and what differentiates agentic systems from traditional models. It will then move into practical implementation using cutting-edge frameworks and tools, demonstrating how to design agent workflows, orchestrate model capabilities, and integrate external resources. This talk is ideal for researchers, engineers, and practitioners interested in next-generation agentic FM systems that advance beyond static models toward dynamic, autonomous, goal-driven agents.

Title: Towards Federated Agentic AI

Speaker: Guodogn Long, University of Technology Sydney, Australia

Biography
A/Prof. Guodong Long is an Associate Professor at the University of Technology Sydney (UTS), where he leads the Foundation Model and Federated Learning Group (https://www.fmfl.group/). His research focuses on federated learning, trustworthy AI, privacy-preserving and personalized intelligence, and on-device intelligence cooperating with pre-trained foundation models such as large language models.

This talk introduces the progression from classical AI models and agents to the emerging paradigm of Federated Agentic AI. It highlights the motivation for extending federated learning into federated agentic systems, outlines the key technical challenges in enabling decentralized, autonomous agents, and presents emerging frameworks and workflows for building privacy-preserving, adaptive, and collaborative agentic AI.