Knowledge Extraction via Learning and Causality Analysis for Event Detection for Resilience Emergency Response

This research will contribute to, and provide the opportunity for renovating the way existing disaster resilience program currently practiced, turning into evidence-based and precision-based approaches. Additionally, the developed data-driven algorithms in the research will be modelled emphasizing the socio-economic aspects of the consequences and cascading losses by allowing our system to adapt according to the community based variables and the dynamics of the disasters. The research objectives are:

• A few-sample based learning algorithm for finding new events in timely manner. The algorithm is also featured with situational awareness, trust, and relevance measures which are valuable for enriching surrounding context of the events.
• A spatialtemporal causality analysis and inter-event reasoning techniques which allow linkages of multiple events in time, space, and between events, which can greatly contribute towards research in transfer learning.
• Advancing transfer learning and graph neural network algorithms for disaster resilience, to fill important knowledge gaps at the intersection of the machine learning and disaster resilience.

Researcher