Human-in-the-Loop Explainable AI : A Case Study in Flood Response (Core Science Keynote)

In this talk I will briefly touch on recent advances in human-in-the-loop explainable AI. I will then describe specific work on how such ideas can be leveraged during crises for crowdsourced remote sensing and flood mapping during hurricane response. As part of this effort I will also spend some time discussing the importance of handling anomalies and biases inherent to such data and how effectively modeling such information can lead to better performance and outcomes.