The Applied Intelligence Research Centre has a diverse team.
The use of Neural Networks (NN) in industry and research has increased and with this so it has the amount of energy consumed and computational power required for data processing and modelling. This trend has result in a great increase in carbon emissions and the amount of power required for training networks. One way to avoid this problem is by making the training of DNN more efficient and this area is the focus of my research project. Currently I am working on measuring of the efficiency (Watts, Number of memory reads, writes and FLOPs on the GPU) of Bayesian Convolutional Neural Networks, and of Neural Ordinary Differential Equations (Implicit Architectures) , I am also looking into Topological Data Analysis methods to search for efficient architectures.