AI to guide Science & Geometric Deep Learning of Surfaces

On March 8 the NWO Domain Board Science has approved nineteen grant applications in the Open Competition Domain Science-M programme, two of which for projects within EEMCS. Sebastijan Dumancic will research how Artificial Intelligence (AI) can help scientists with discovering explanations for natural phenomena. Klaus Hildebrandt will work on geometric convolutional neural networks to see how they can help solve difficult problems in surface analysis and processing. Although already existing, these methods are still under development and need to be further developed before they can be widely used in practice. 

Multigrid methods for learning from 3D-data

Klaus Hildebrandt's project is inspired by the successful application of convolutional neural networks (CNNs) in image analysis and the growing availability of 3D data. In recent years, researchers have developed CNNs for triangle meshes and point clouds, which have shown potential in solving complex problems related to surface analysis and processing. His project aims to develop novel network structures and training techniques for geometric deep learning using triangle meshes and point clouds, specifically focusing on geometric multigrid methods and multiscale training. The ultimate goal of the project is to enhance the practical use of geometric CNNs in surface analysis and processing.

AI as a guide for scientists

Richard Feynman famously said "What I cannot create, I do not understand". This statement emphasizes the importance of gaining a comprehensive understanding of a concept by starting from scratch, using fundamental principles and building blocks. Scientists and engineers adopt this principle when they simplify natural phenomena and create mathematical equations to explain them. Scientists and engineers follow this principle when they distil natural phenomena into concise mathematic equations and build test prototypes of complex machines. However, this process can be labor-intensive and time-consuming, leading to the question of whether automation can assist in this endeavor.

Dumancic and his team want to explore the possibility of using artificial intelligence (AI) to help uncover explanations for the natural phenomena that surround us. The goal is to create an AI system that can help scientists and engineers gain a deeper understanding of complex systems by automating the process of deriving equations and building test prototypes. Such a system would significantly accelerate the process of scientific discovery and engineering innovation.