Hierarchical information processing in the primate visual cortex. This project aims to understand how visual information is transformed across hierarchical levels in the brain. Neuroscientists have long recognised that the visual cortex can be conceptualised as a hierarchical processing network. This became apparent when learning algorithms based on hierarchical networks ("deep learning") changed artificial intelligence. This project will combine high-throughput electrophysiology with analytical ....Hierarchical information processing in the primate visual cortex. This project aims to understand how visual information is transformed across hierarchical levels in the brain. Neuroscientists have long recognised that the visual cortex can be conceptualised as a hierarchical processing network. This became apparent when learning algorithms based on hierarchical networks ("deep learning") changed artificial intelligence. This project will combine high-throughput electrophysiology with analytical tools adopted from deep learning. By explaining the physiological properties of higher-level neurons in terms of hierarchical networks, the project expects to address long standing questions in neuroscience, and provide insights on biological hierarchical computation.Read moreRead less
Mathematical modelling of information flow in social networks. This proposal aims to develop new mathematical and statistical methods to understand information flow in social networks. By using novel information theoretic techniques, it will create new methods to characterise social information flow in social networks. These tools will allow derivation of fundamental limits of predictability for AI methods applied to digital data. New mathematics of information flow will produce insights into so ....Mathematical modelling of information flow in social networks. This proposal aims to develop new mathematical and statistical methods to understand information flow in social networks. By using novel information theoretic techniques, it will create new methods to characterise social information flow in social networks. These tools will allow derivation of fundamental limits of predictability for AI methods applied to digital data. New mathematics of information flow will produce insights into social influence in online social networks. Benefits include: better understanding of how echo chambers may form in social networks, predictive models for how misinformation can spread online such as during an emergency, and a framework for intercomparison of AI methods applied to digital data on individuals. Read moreRead less