Workshops and Tutorials

Date: Tuesday, 1 December 2026
Location: University of Technology Sydney (UTS), Building 11, Sydney, Australia
Time zone: Australian Eastern Daylight Time (AEDT)

Schedule

Tuesday, 1 December 2026

Time (AEDT) Session Location
09:00–12:30 Workshop 1: Towards Trustworthy Visual and Multimodal Agents — Morning Session CB11.B1.101
09:00–13:00 Workshop 2: Vision-Language and Agentic AI Modeling in Image-Guided Theranostics CB11.B3.102
12:30–14:00 Lunch Break UTS Building 11
14:00–17:30 Workshop 1: Towards Trustworthy Visual and Multimodal Agents — Afternoon Session CB11.B1.101
14:00–17:30 Tutorial 1: Autonomous Drift Learning: Toward Self-Evolving Intelligence in a Dynamic World CB11.B3.102
14:00–17:30 Tutorial 2: Visual Agents for Computer Use: Perception, Reasoning, and Safety CB11.09.118 Common Room

Tea breaks are included within the session times.

Workshops

Workshop 1: Towards Trustworthy Visual and Multimodal Agents

Organisers: Feng Liu, Yadan Luo, Chengkai Huang, Jing Du, and Zesheng Ye

Format: Full-day workshop

Time: 09:00–12:30 and 14:00–17:30

Room: CB11.B1.101

Abstract

As multimodal foundation models become the foundation of increasingly capable visual agents, these systems must interpret visual inputs, connect observations to supporting evidence, reason across long interactions, and use external tools in changing environments. This workshop examines how to make such agents dependable. It brings together work on robustness, safety, fairness, privacy, calibration, provenance, hallucination detection, grounding, red-teaming, and evaluation of complete agent trajectories. The discussions also consider distribution shifts and high-stakes deployment, with an emphasis on developing more consistent evaluation approaches and practical reliability standards.

Workshop 2: Vision-Language and Agentic AI Modeling in Image-Guided Theranostics

Organisers: Olatunji (Ola) Omisore, Anwaar Ul-Haq, and Jack Yang

Format: Half-day workshop

Time: 09:00–13:00

Room: CB11.B3.102

Workshop website: Vision-Language and Agentic AI in Image-Guided Theranostics

Abstract

This workshop explores how vision-language technologies, large language models, and agentic AI may support image-guided medical treatment and intervention. Sessions address medical-image and robotic foundation models, retrieval-supported clinical decision-making, multimodal approaches to procedural guidance, surgical scene interpretation, workflow and action recognition, three-dimensional reconstruction, spatial reasoning, and collaboration between clinicians and AI systems. Bringing technical and clinical perspectives together, the workshop aims to surface open research problems and encourage reproducible approaches to safe, clinically useful intelligent healthcare.

Invited Speakers

Panel Discussion

Defining the Next Generation of Image-Guided Theranostics with Vision-Language and Agentic AI

Panellists:

Tutorials

Tutorial 1: Autonomous Drift Learning: Toward Self-Evolving Intelligence in a Dynamic World

Presenters: Wei Duan, Xiaoyu Yang, and En Yu

Format: Half-day tutorial

Time: 14:00–17:30

Room: CB11.B3.102

Abstract

This tutorial provides an introduction to autonomous drift learning, focusing on systems that must continue learning as their environments and underlying data change. It connects three lines of research: detecting and adapting to shifts in streaming data, maintaining continual adaptability in large-scale models, and developing agents that can evolve through interaction. Participants will gain an overview of the key concepts, methodological approaches, and practical challenges involved in building AI that remains adaptive and reliable under non-stationary conditions.

Tutorial 2: Visual Agents for Computer Use: Perception, Reasoning, and Safety

Presenters: Zijing Shi and Chen Liu

Format: Half-day tutorial

Time: 14:00–17:30

Room: CB11.09.118 Common Room

Abstract

This tutorial examines agents that operate computer interfaces through visual observations and actions. It focuses on three connected problems: perceiving and grounding elements in graphical user interfaces, reasoning and planning over extended sequences of interactions, and ensuring safety and dependable behaviour. The programme surveys common system designs, public benchmarks, evaluation practices, and failure patterns, with practical guidance for researchers building and assessing computer-use agents.