Sho Okazaki
Researcher in Causal and Neuro-Symbolic AI for Intelligent Asset Management
I am a researcher developing AI methods for understanding, predicting, and managing failures in complex cyber-physical systems. My research integrates neuro-symbolic AI, causal reasoning, machine learning, and multimodal sensing to support anomaly detection, root-cause analysis, risk assessment, and maintenance decision-making. Working across domains including manufacturing systems and railway infrastructure, I aim to bridge data-driven intelligence and engineering knowledge to improve the resilience, reliability, and sustainability of critical assets.
News
- 2025-10-01
Awarded Nakajima Foundation Scholarship and Chadwick Scholarship.
- 2025-10-01
Joined UCL as a Doctoral Researcher.
- 2025-03-24
Master's thesis was awarded the Excellence Award by Department of Precision Engineering at the University of Tokyo.
- 2024-11-01
Joined the Committee of Young Researcher Association for Knowledge Graph in Japan.
Featured Publications
View allA spatio-temporal anomaly detection system to support understanding of abnormal phenomena in automated manufacturing lines
Okazaki, S., Kaminishi, K., Wang, Y., Fujiu, T., Nakata, Y., Hamamoto, S., Yokose, K., Hara, T., Umeda, Y., Ota, J.
Machine Learning Approach to Redefining Risk in Railway Drainage Systems
Okazaki, S., Herrera, M., Sasidharan, M., McNaughton, J., Raja, J., Parlikad, A. K.
FBS model-based maintenance record accumulation for failure-cause inference in manufacturing systems
Fujiu, T., Okazaki, S., Kaminishi, K., Nakata, Y., Hamamoto, S., Yokose, K., Hara, T., Umeda, Y., Ota, J.
Recent Writing
View allCurrent Projects
View allWhole-System Approach to Railway Asset Management
A research project that investigates the interdependencies among railway assets and establishes a whole-system approach to railway asset management.
Knowledge-Driven Anomaly Detection and Diagnosis in Manufacturing Systems
A research project that evolved from fault-cause diagnosis into maintenance-record modeling, graph learning, and upstream anomaly detection.