AI RESEARCH

Computer-Aided Design Generation by Cascaded Discrete Diffusion Model

arXiv CS.CV

ArXi:2605.05031v1 Announce Type: new Recent deep learning approaches seek to automate CAD creation by representing a model as a sequence of discrete commands and parameters, and then generating them using autoregressive models or continuous diffusion operating in Euclidean embedding space. However, continuous diffusion perturbs representations in a continuous Euclidean domain that does not reflect the inherently discrete and heterogeneous nature of CAD tokens, often producing perturbed representations that map to semantically invalid symbols.