Research map / Graph learning / GNN architectures
Penguin: Parallel-Packed Homomorphic Encryption for Fast Graph Convolutional Network Inference
Ran Ran, Nuo Xu, Tao Liu, Wei Wang, Gang Quan, Wujie Wen
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 |
|---|---|---|---|
| two-dimension packing (16,256) instead of feature-only packing (1433,1) (ablation) | Penguin · Cora | latency: 7018.51 -> 678.03 | Improved |
| packing (32,128) instead of (1433,1) (ablation) | Penguin · Cora | latency: 7018.51 -> 871.15 | Improved |
| packing (64,64) instead of (1433,1) (ablation) | Penguin · Cora | latency: 7018.51 -> 1650.28 | Improved |
| two-dimension packing reduces rotations (ablation) | Penguin · Cora | rotations: 1048K -> 9.7K | Improved |
| packing (64,64) instead of (19717,1) (ablation) | Penguin · PubMed | latency: 44586.03 -> 80424.70 | Hurt |
| packing (16,256) instead of (19717,1) (ablation) | Penguin · PubMed | latency: 44586.03 -> 30906.28663 | Improved |
| packing (32,128) instead of (19717,1) (ablation) | Penguin · PubMed | latency: 44586.03 -> 40474.11 | Improved |
| add interleaved assembling to Penguin(32,128) (ablation) | Penguin(32,128) · Cora | latency: 871.15 -> 660.67 | Improved |
| add interleaved assembling to Penguin(32,128) (ablation) | Penguin(32,128) · Citeseer | latency: 1225.05 -> 928.10 | Improved |
| add interleaved assembling to Penguin(32,128) (ablation) | Penguin(32,128) · PubMed | latency: 40474.11 -> 30522.43 | Improved |
| add interleaved assembling to Penguin(64,64) (ablation) | Penguin(64,64) · Cora | latency: 1650.28 -> 693.13 | Improved |
| add interleaved assembling to Penguin(32,128) (ablation) | Penguin(32,128) · Cora | CMult count: 124K -> 9.3K | Improved |
Closest papers by meaning
- FicGCN: Unveiling the Homomorphic Encryption Efficiency from Irregular Graph Convolutional Networks (ICML 2025 · similarity 0.91)
- PrivCirNet: Efficient Private Inference via Block Circulant Transformation (NeurIPS 2024 · similarity 0.85)
- DCT-CryptoNets: Scaling Private Inference in the Frequency Domain (ICLR 2025 · similarity 0.85)
- Converting Transformers to Polynomial Form for Secure Inference Over Homomorphic Encryption (ICML 2024 · similarity 0.85)
- GRAIN: Exact Graph Reconstruction from Gradients (ICLR 2025 · similarity 0.85)
- GraphIP–Bench: How Hard Is It to Steal a Graph Neural Network, and Can We Stop It? (NeurIPS 2026 · similarity 0.85)
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.
