Presentation Date:
Location:
This was an invited talk for the À¶Ý®ÊÓƵ Institute for Complexity and Innovation (WICI) seminar series. The talk was recorded and can be watched from WICI's website here.
Abstract:
Recent advancesÌýin Artificial Intelligence and Machine LearningÌý(AI/ML) allowÌýus to learn predictive models and control policies for larger, more complex systems than ever before.ÌýHowever, someÌýimportant real world domains such as forest wildfire spread, flooding​ andÌýmedical imaging present a particular challenge. They containÌýspatially spreadingÌýprocesses(SSP)Ìýwhere some local features change over time based on proximity inÌýspace. This talk will present newÌýapproaches to​ learning for SSPs usingÌýDeep Learning and Reinforcement Learning such as learningÌýa model ofÌýwildfireÌýspread from satellite images as if the wildfire were an agent making decisions about where to move next.ÌýThis approach could lead to learning models which are more interpretable and aidÌýdomain experts in creating rich, agent-based models based onÌýdata.
Ìý
