📝 Summary
Prof. Ts. Dr. Ali Selamat presented an AI-driven approach to disaster management at the 2026 ASEAN Science and Education Global Expansion Forum in Shenzhen, China, focusing on leveraging Generative Adversarial Networks (GAN) to enhance flood forecasting accuracy. The approach aims to address data limitations in early warning systems by reconstructing missing data, improving preparedness and emergency response. The presentation highlighted the potential for Malaysia to accelerate AI adoption in crisis management systems.
SHENZHEN, 10 Apr – Universiti Teknologi Malaysia (UTM) continues to strengthen its role as a driver of Innovating Sustainable Solutions as its Deputy Vice Chancellor (Student Affairs and Alumni), Prof. Ts. Dr. Ali Selamat, who is recognised among the World’s Top 2% Scientists for Career-Long Achievement 2024 by Elsevier B.V. and Stanford University, was invited to present a high-impact artificial intelligence (AI)-driven approach to disaster management at Shenzhen Virtual University Park (SZVUP), China, in conjunction with the 2026 ASEAN Science and Education Global Expansion Forum.
The presentation, titled “Artificial Intelligence for Disaster Management: Generative Adversarial Network (GAN)-based Missing Data Imputation for Time-Series Rainfall in Flood Prediction,” focused on leveraging artificial intelligence (AI) to enhance the accuracy of flood forecasting, particularly in addressing data limitations that remain a critical challenge in early warning systems.
According to him, reliable and complete data is fundamental to effective predictive modelling. However, disruptions in telemetry systems during extreme weather conditions often result in critical data loss, creating “blind spots” that significantly affect the accuracy of disaster forecasting.
“AI is no longer an option, but a strategic necessity in ensuring that disaster prediction systems are more responsive and resilient. By enabling the reconstruction of missing data, AI enhances preparedness and accelerates emergency response,” he said.
During the presentation, he and his research team introduced an innovative framework integrating key technologies, including Generative Adversarial Networks (GAN) for synthetic data generation, Temporal Convolutional Networks (TCN) for temporal pattern analysis, and Transformer models for real-time data processing in flood classification.
The approach was validated through a case study at Sungai Kuantan, Pahang, demonstrating promising results. The developed models improved forecasting accuracy, reduced prediction errors compared to conventional methods, and achieved high accuracy levels in flood detection through satellite image analysis.
He further noted that advanced economies such as China, Japan, and Korea have already integrated AI extensively into crisis management systems. Malaysia, he added, holds strong potential to accelerate AI adoption through strengthened policy frameworks, resource mobilisation, capacity development, and enhanced digital literacy among the public.
“While we cannot prevent monsoon events, empowering AI to identify and close data ‘blind spots’ ensures that decisions are informed, timely, and comprehensive,” he added.
The presentation was delivered as part of his official working visit to Shenzhen Virtual University Park, opening avenues for potential international research collaborations in artificial intelligence and disaster management. This initiative is expected to serve as a catalyst for the development of more holistic and high-impact AI-driven solutions, in line with UTM’s aspiration to lead technological innovation for the betterment of global society.
