Climate change is causing wide-ranging impacts on society that will only become even more amplified as the phenomenon progresses. One such impact is the prevalence of natural disasters, which increase in frequency and intensity due to the warming globe. In order to respond to disasters, we envision the development of interconnected AI-driven systems that detect damage and enable near-instantaneous pipelines for disaster relief. These technologies utilize computer vision to assess the severity of damage and aid in the timely and equitable allocation of resources. While currently AI models are being developed, the deployment process is complex, especially in the context of IoT. In this session, we discuss a pipeline for creating such systems by first training convolutional neural networks (CNNs) both on multitemporal earth observation data (satellite imagery) and on social media data. Both of these sources of data are key assets in assessing impacts of natural disasters (from above and on the ground), and can enable the systems that are deployed in a mobile setting. We'll briefly examine the efficacy and efficiency of the aforementioned deep learning models/algorithms in completing the task at hand. Finally, we discuss the deployment process, whereby ethics, interpretability, and accessibility considerations must be taken into account. These apps will aid in the allocation of resources and personnel in these devastating events and help save lives, property, time, and economic resources.
Mobile Artificial Intelligence-Driven Applications for Natural Disaster Relief
Thomas Chen is a machine learning researcher from the USA that is passionate about machine learning, computer vision, and artificial intelligence. He is highly involved in applying ML and AI to real-world issues that face society (e.g. deep learning-based computer vision for damage assessment post-natural disaster). He has been an invited speaker at conferences like the IEEE Conference on Technologies for Sustainability, American Geophysical Union Fall Meeting, and the Energy Anthropology Network. He is considered a leading expert in the area of machine learning-driven earth observation (EO) applications.