Research map / Graph learning / Knowledge and temporal graphs
Asynchronous Event Processing with Local-Shift Graph Convolutional Network
Linhui Sun, Yifan Zhang, Jian Cheng, Hanqing Lu
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 spatial graph convolution with local-shift (ablation) | LSNet · N-Cars | MFLOPs: 940.9 -> 105.1 | Improved |
| replace spatial graph convolution with local-shift (ablation) | LSNet · N-Cars | accuracy: 95.45 -> 95.51 | Improved |
| add local spatio-temporal attention (ablation) | LSNet with local-shift · N-Cars | accuracy: 95.51 -> 95.78 | Improved |
| add global attention (ablation) | LSNet with local-shift and local attention · N-Cars | accuracy: 95.78 -> 96.10 | Improved |
| replace NIPooling with random pooling (ablation) | LSNet · N-Cars | accuracy: 95.81 vs 96.10 | Hurt |
| replace NIPooling with FPS (ablation) | LSNet · N-Cars | average pooling time: 85.81 vs 0.11 | Hurt |
| replace NIPooling with FPS (ablation) | LSNet · N-Cars | accuracy: 95.88 vs 96.10 | Hurt |
| use asynchronous event processing instead of recomputing all graph activations | LSNet · MNIST-DVS | MFLOPs/event: 112 -> 3.3 | Improved |
| use asynchronous event processing instead of recomputing all graph activations | LSNet · N-Cars | MFLOPs/event: 117.2 -> 2.3 | Improved |
| use asynchronous event processing instead of recomputing all graph activations | LSNet · CIFAR10-DVS | MFLOPs/event: 251.2 -> 2.0 | Improved |
| set minimum neighbor count Kmin to 4 (ablation) | ALSNet · N-Cars | accuracy: 96.10 -> 95.74 | Hurt |
| set minimum neighbor count Kmin to 16 (ablation) | ALSNet · N-Cars | accuracy: 96.10 -> 95.95 | Hurt |
Closest papers by meaning
- AEGNN: Asynchronous Event-Based Graph Neural Networks (CVPR 2022 · similarity 0.91)
- Graph Neural Network Combining Event Stream and Periodic Aggregation for Low-Latency Event-based Vision (CVPR 2025 · similarity 0.89)
- Asynchronous Collaborative Graph Representation for Frames and Events (CVPR 2025 · similarity 0.86)
- Leveraging Asynchronous Spiking Neural Networks for Ultra Efficient Event-Based Visual Processing (AAAI 2025 · similarity 0.86)
- ALERT-Transformer: Bridging Asynchronous and Synchronous Machine Learning for Real-Time Event-based Spatio-Temporal Data (ICML 2024 · similarity 0.85)
- Are Spatial-Temporal Graph Convolution Networks for Human Action Recognition Over-Parameterized? (CVPR 2025 · 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.
