Thomas Nowotny

You will develop novel active AI algorithms that are inspired by the rapid and robust learning of insects within the £1.2m EPSRC International Centre to Centre Collaboration project: “ActiveAI: active learning and selective attention for rapid, robust and efficient AI.” and will work in collaboration with the University of Sheffield and world-leading neuroscientists in Australia. Your primary role will be to develop a new class of ActiveAI controllers for problems in which insects excel but deep learning methods struggle. These problems have one or more of the following characteristics: (i) learning must occur rapidly, (ii) learning samples are few or costly, (iii) computational resources are limited, and (iv) the learning problem changes over time. Insects deal with such complex tasks robustly despite limited computational power because learning is an active process emerging from the interaction of evolved brains, bodies and behaviours. Through a virtuous cycle of modelling and experiments, you will develop insect-inspired models, in which behavioural strategies and specialised sensors actively structure sensory input while selective attention drives learning to the most salient information. The cycle of modelling and experiments will be achieved through field work in both Sussex and Australia.