SketchDNN: Joint Continuous-Discrete Diffusion for CAD Sketch Generation

2025-07-15

AI Summary

This work addresses two key challenges in CAD sketch generation: the heterogeneous coexistence of continuous parameters and discrete category labels, and the permutation invariance of geometric primitives. We propose a unified continuous-discrete joint diffusion model. Its core innovation is the Gaussian-Softmax diffusion mechanism: Gaussian noise is applied to discrete-category logits, followed by a softmax projection onto the probability simplex—enabling differentiable, trainable discrete modeling that naturally couples with continuous-parameter diffusion. To our knowledge, this is the first diffusion-based approach to jointly model both primitive types (discrete) and their geometric parameters (continuous), ensuring semantic consistency and structural invariance. On the SketchGraphs dataset, our method achieves state-of-the-art performance: FID improves significantly from 16.04 to 7.80, and negative log-likelihood decreases from 84.8 to 81.33.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Search and Optimization: Mixed Discrete/Continuous Search
• Constraint Satisfaction and Optimization: Mixed Discrete/Continuous Optimization

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

We present SketchDNN, a generative model for synthesizing CAD sketches that jointly models both continuous parameters and discrete class labels through a unified continuous-discrete diffusion process. Our core innovation is Gaussian-Softmax diffusion, where logits perturbed with Gaussian noise are projected onto the probability simplex via a softmax transformation, facilitating blended class labels for discrete variables. This formulation addresses 2 key challenges, namely, the heterogeneity of primitive parameterizations and the permutation invariance of primitives in CAD sketches. Our approach significantly improves generation quality, reducing Fréchet Inception Distance (FID) from 16.04 to 7.80 and negative log-likelihood (NLL) from 84.8 to 81.33, establishing a new state-of-the-art in CAD sketch generation on the SketchGraphs dataset.

Problem

• Modeling continuous and discrete CAD sketch parameters jointly
• Addressing heterogeneity and permutation invariance in CAD sketches
• Improving generation quality with Gaussian-Softmax diffusion

Innovation

• Unified continuous-discrete diffusion process
• Gaussian-Softmax diffusion for blended labels
• Handles parameter heterogeneity and permutation invariance