Скачать или смотреть Iain Couzin: Employing Immersive Virtual Reality and Machine Learning ...

Full Title: Employing Immersive Virtual Reality and Machine Learning to Reveal Geometric Principles of Individual and Collective Decision-Making

Iain Couzin is Director of the Max Planck Institute of Animal Behavior, Department of Collective Behaviour and the Chair of Biodiversity and Collective Behaviour at the University of Konstanz, Germany. This talk was part of the colloquium of the Cluster of Excellence "Machine Learning: New Perspectives for Science".

Abstract: I will discuss our development and application of new machine learning technologies for automated tracking and identification of unmarked individuals, computational sensory reconstruction, bio-mimetic robotics, and ‘holographic’ virtual reality (VR) for animals, and demonstrate how these quantitative methodologies provide new insights into individual and collective sensing and decision-making. For example, they allow us to reconstruct (automatically) the dynamic, time-varying sensory networks in large animal collectives, to identify at any instant in time the most socially-influential individuals within groups, and to predict cascades of social contagion (behavioural change) before they actually occur. Furthermore, such methodologies enable us to study phenomena across multiple scales of biological organisation, from neural interactions to individual and collectives decision-making. Specifically, we have recently investigated how animals choose among spatially-discrete options, a central challenge in their lives. Employing an integrated theoretical and experimental approach (using immersive VR), with both invertebrate and vertebrate models—the fruit fly, desert locust and zebrafish—we find that the brain spontaneously reduces multi-choice decisions into a series of abrupt (critical) binary decisions in space-time, a process that repeats until only one option—the one ultimately selected by the individual—remains. This mechanism facilitates highly effective decision-making, and is shown to be robust both to the number of options available, and to context, such as whether options are static (e.g. refuges) or mobile (e.g. other animals). In addition, we find evidence that the same geometric principles of decision-making occur across scales of biological organisation, from neural dynamics to animal collectives, suggesting they are fundamental features of spatiotemporal computation.