IEEE CISOSE 2026 will feature three half-day tutorials (approximately 3 hours each), delivered in person at Fukuoka Institute of Technology. Session times will be announced in the final program.
INTRODUCTORY SURVEY · HALF DAY (~3 HOURS) · ENGLISH
Tutorial 1 — From Test Automation to Agentic Test Automation for Smart Systems
Presenter: Jerry Zeyu Gao (San Jose State University, California, USA)
According to Global Market Insights, the automation-testing market is projected to grow from US$15B in 2020 to more than US$40B by 2027, and the rapid rise of AI-powered systems is driving strong demand for AI system test automation. This survey tutorial reviews AI system testing and automation — test modeling and analysis, test generation and augmentation, smart test validation, and the use of LLMs in test automation — then examines the major problems and needs, including intelligence-quality validation and test-coverage evaluation. It closes with a roadmap that transforms test automation into agentic AI test automation across three smart-system domains: AI systems, smart robots, and autonomous vehicles. Intended for AI system test engineers, managers, researchers, students, and industry practitioners.
Schedule: TBA.
HANDS-ON / TOOL DEMO · HALF DAY (~3 HOURS) · ENGLISH
Tutorial 2 — Building and Verifying AI Systems with Agentic AI
Presenters: Chia-Kai Chang (National Central University, Taiwan); Kuei-Hao Li (National Tsing Hua University, Taiwan); Teng-Yok Lee (Mitsubishi Electric, Japan)
Generative and agentic AI have collapsed the cost of building real-world, multi-user AI systems — but not the cost of understanding or trusting them. This hands-on tutorial teaches a domain-agnostic method in three repeating moves: build a system with spec-driven, agentic AI development; make it legible with data visualization; and use that visualization to test it. Attendees leave with a layered AI-testing rubric (code, pipeline, behavior, guardrail, governance, and drift), a runnable LLM-as-Judge harness with versioned prompts and cross-model reproducibility, visualization methods (CAM-style contribution maps and visual anomaly detection), and a redacted, IRB-governed data-handling template. A production AI education platform serves as the running example, but the method is the takeaway — not a tour of one platform.
Tutorial page: https://uedu.tw/cisose-2026-tutorial/
Schedule: TBA.
INTRODUCTORY / FOUNDATIONS · HALF DAY (~3 HOURS) · ENGLISH
Tutorial 3 — Foundations of Physical AI: Sensing, Calibration, Communication, and Data Validation
Presenter: Winncy Du (San José State University, California, USA)
Physical AI systems — autonomous robots, drones, intelligent manufacturing systems, and cyber-physical infrastructures — depend fundamentally on the quality and reliability of sensor data. While recent advances in AI have focused heavily on machine learning models, the performance of Physical AI ultimately begins with sensing, data acquisition, communication, and validation: inaccurate measurements, poor calibration, communication failures, and low-quality data can significantly degrade AI-driven decision-making. This tutorial provides a comprehensive introduction to the foundational technologies that enable trustworthy Physical AI — sensor selection and characterization, calibration and uncertainty analysis, sensor networking and communication protocols, data quality assessment, and validation techniques for real-world deployments — along with sensor fusion, edge intelligence, and emerging approaches for integrating AI with sensing systems. Through practical examples and case studies from robotics, unmanned aerial systems, smart manufacturing, and IoT applications, attendees will gain insights into designing end-to-end sensing architectures that support reliable AI operations. Intended for researchers, practitioners, and graduate students interested in trustworthy Physical AI and the critical role of sensing and data quality in AI-enabled cyber-physical environments.
Schedule: TBA.
All tutorials are presented in person at Fukuoka Institute of Technology, Japan; remote or hybrid delivery is not permitted.