B Rep

B-rep (Boundary Representation) is a fundamental data structure in computer-aided design (CAD) representing 3D solids as a collection of faces, edges, and vertices. Current research focuses on developing efficient methods for generating and manipulating B-rep models using machine learning, particularly employing diffusion models and neural implicit representations like neural halfspace representations (NH-Rep). These advancements aim to improve the creation, manipulation, and analysis of complex CAD geometries, impacting fields like automated design, CAD software development, and digital fabrication. The ability to learn from unlabeled data is also a key area of focus, addressing the scarcity of labeled CAD datasets.