Intelligent Earth Seminar: Andrew Markham (Oxford)

From Sensors to Species: Embedded Machine Learning for Wildlife Monitoring

13 November 14:00
Earth Sciences: Lecture Theatre

Andrew Markham is a professor of Computer Science at the University of Oxford, UK. He received his BSc (Hons) (2004) and PhD (2008) both from the University of Cape Town, South Africa in Electrical Engineering. He has published over 200 papers and works in the broad area of cyberphysical systems, particularly focussing on multimodal sensing and efficiency/accuracy/robustness tradeoffs. His research interests are on the application of sensing, systems, and signal processing to increased sustainability, conservation, and autonomy. 

From Sensors to Species: Embedded Machine Learning for Wildlife Monitoring

Key to better conservation of natural systems is the ability to understand how animals behave within their environments at various levels of granularity, from the species to the individual. In this talk, I outline a systems approach to exploiting machine learning at various levels of computation (e.g. on device vs in cloud) to efficiently sense, communicate, and process information for animal tracking and monitoring. Beyond the conventional goal of improved accuracy, I also look at how we can engineer these devices through hardware-software codesign to acquire richer information for longer and at a lower cost. Drawing from real-world deployments across four continents I show how embedded machine learning has the potential to provide new insights into the wild.