Research map / Graph learning / GNN architectures
Deep Message Passing on Sets
Yifeng Shi, Junier Oliva, Marc Niethammer
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 |
|---|---|---|---|
| replace the learned latent graph with a uniform graph (ablation) | V-DMPS · Counting unique characters | test accuracy: 0.6446 ± 0.0174 -> 0.1357 ± 0.0000 | Hurt |
| replace the learned latent graph with a uniform graph (ablation) | D-DMPS w/ LDC · Counting unique characters | test accuracy: 0.6674 ± 0.0080 -> 0.4661 ± 0.0085 | Hurt |
| add set-residual blocks (ablation) | DMPS · Counting unique characters | test accuracy: 0.6446 ± 0.0174 -> 0.6600 ± 0.0103 | Improved |
| add set-denoising blocks with fixed diffusion coefficient (ablation) | DMPS · Counting unique characters | test accuracy: 0.6446 ± 0.0174 -> 0.6748 ± 0.0120 | Improved |
| add set-denoising blocks with learnable diffusion coefficient (ablation) | DMPS · Counting unique characters | test accuracy: 0.6446 ± 0.0174 -> 0.6674 ± 0.0080 | Improved |
| set the diffusion coefficient too high or too low (ablation) | D-DMPS w/ FDC · ModelNet40, 100 points | test accuracy | Hurt |
| add set-residual blocks (ablation) | DMPS · ModelNet40, 100 points | test accuracy: 0.8367 ± 0.0047 -> 0.8475 ± 0.0036 | Improved |
| add set-denoising blocks (ablation) | DMPS · ModelNet40, 100 points | test accuracy: 0.8367 ± 0.0047 -> 0.8571 ± 0.0062 | Improved |
| add set-residual blocks (ablation) | DMPS · ModelNet40, 1000 points | test accuracy: 0.8751 ± 0.0029 -> 0.8935 ± 0.0016 | Improved |
| add set-denoising blocks (ablation) | DMPS · ModelNet40, 1000 points | test accuracy: 0.8751 ± 0.0029 -> 0.8798 ± 0.0020 | Improved |
| add set-residual blocks (ablation) | DMPS · Breast Cancer | test accuracy: 0.800 ± 0.023 -> 0.818 ± 0.029 | Improved |
| add set-denoising blocks (ablation) | DMPS · Breast Cancer | test accuracy: 0.800 ± 0.023 -> 0.846 ± 0.019 | Improved |
Closest papers by meaning
- Adaptive Message Passing: A General Framework to Mitigate Oversmoothing, Oversquashing, and Underreaching (ICML 2025 · similarity 0.85)
- Geometric Generative Modeling with Noise-Conditioned Graph Networks (ICML 2025 · similarity 0.85)
- SDMG: Smoothing Your Diffusion Models for Powerful Graph Representation Learning (ICML 2025 · similarity 0.84)
- Outlier-Aware Post-Training Quantization for Discrete Graph Diffusion Models (ICML 2025 · similarity 0.84)
- Is Noise Conditioning Necessary? A Unified Theory of Unconditional Graph Diffusion Models (NeurIPS 2025 · similarity 0.84)
- Demystifying Topological Message-Passing with Relational Structures: A Case Study on Oversquashing in Simplicial Message-Passing (ICLR 2025 · similarity 0.84)
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.
