Deep learning
The research group works on the use of deep learning to approximate problems in very complex geometries, using artificial intelligence as a tool to build reduced models. The activity is aimed both at the introduction of new architectures to improve the approximation capabilities of neural networks, and at the use of explainable AI strategies to derive useful information to reduce the complexity of the computational domain. The group also works on the integration between the most innovative machine learning techniques and models capable of simulating highly complex physical events, such as solar flares.