Research map / Graph learning / GNN architectures
Graph Transformer GANs With Graph Masked Modeling for Architectural Layout Generation
Hao Tang, Ling Shao, Nicu Sebe, Luc Van Gool
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use graph transformer generator instead of HouseGAN generator (ablation) | HouseGAN · house layout (10-12) | FID: 11.6 -> 10.3 | Improved |
| use GTGAN discriminator instead of HouseGAN discriminator (ablation) | HouseGAN · house layout (10-12) | FID: 11.6 -> 9.5 | Improved |
| use both GTGAN generator and discriminator (ablation) | HouseGAN · house layout (10-12) | FID: 11.6 -> 8.2 | Improved |
| add node-neighbor attention (NNA) (ablation) | GTGAN · house layout (10-12) | FID: 9.4 -> 8.2 | Improved |
| add connected-node attention (CNA) (ablation) | GTGAN · house layout (10-12) | FID: 9.7 -> 8.2 | Improved |
| add graph modeling block (GMB) (ablation) | GTGAN · house layout (10-12) | FID: 9.5 -> 8.2 | Improved |
| use Conv-MPN layers instead of Transformer layers (ablation) | GTGAN · house layout (10-12) | FID: 8.9 -> 8.2 | Improved |
| add graph-based cycle-consistency loss (ablation) | GTGAN · house layout (10-12) | FID: 8.2 -> 7.5 | Improved |
| add graph node masking pre-training (ablation) | GTGAN + cycle loss · house layout (10-12) | FID: 7.5 -> 5.9 | Improved |
| add graph edge masking pre-training (ablation) | GTGAN + cycle loss · house layout (10-12) | FID: 7.5 -> 5.7 | Improved |
| add both node and edge masking (ablation) | GTGAN + cycle loss · house layout (10-12) | FID: 7.5 -> 5.3 | Improved |
| set lambda_1 to 1 rather than 0.2 (ablation) | GTGAN++ · house layout (10-12) | FID: 17.7 -> 7.5 | Improved |
Closest papers by meaning
- Graph Transformer GANs for Graph-Constrained House Generation (CVPR 2023 · similarity 0.97)
- House-GAN: Relational Generative Adversarial Networks for Graph-constrained House Layout Generation (ECCV 2020 · similarity 0.90)
- House-GAN++: Generative Adversarial Layout Refinement Network towards Intelligent Computational Agent for Professional Architects (CVPR 2021 · similarity 0.88)
- GraphGAN: Graph Representation Learning With Generative Adversarial Nets (AAAI 2018 · similarity 0.87)
- Building-GAN: Graph-Conditioned Architectural Volumetric Design Generation (ICCV 2021 · similarity 0.87)
- Structure-Aware Transformer for Graph Representation Learning (ICML 2022 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
