Computer Science PGR Seminar Guest Speaker Paul Best

All are welcome to listen to our speaker Paul Best from the Institute of Language, Communication and Brain in France.

Title: Computational bioacoustics in practice, opportunities and limits of using machine learning to study non-human vocal behaviour.

Abstract: In this talk I'll present various use cases of computational passive bioacoustics, with an emphasis on the possibilities offered by machine learning but also its limits. I will demonstrate some techniques we can use to mitigate constraints of power and annotation demand, then illustrating how complementing acoustic acquisition with other information (behaviour and identity) might be crucial to better understand the communication patterns we observe.

Bio: I did my masters in computer science before specialising in bioacoustics for my PhD, at the University of Toulon. I've worked on various cetacean species including sperm whales, killer whales and fin whales; mainly using machine learning tools for use cases such as collision risk mitigation and communication system structure analyses. My current postdoc project now focuses on acoustic spatialisation to contextualise the vocal interactions of pilot whales and gibbons.