Research map / Architectures and efficiency / Spiking networks
Online Pseudo-Zeroth-Order Training of Neuromorphic Spiking Neural Networks
Mingqing Xiao, Qingyan Meng, Zongpeng Zhang, Di He, Dongsheng Li, Zhouchen Lin
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 random feedback (DFA) with OPZO momentum feedback | OTTT online training · CIFAR-10 | accuracy (%): 79.90 -> 85.74 | Improved |
| replace spatial BP with OPZO | OTTT online training · CIFAR-10 | accuracy (%): 89.99 -> 85.74 | Hurt |
| add local learning to OPZO | OPZO · CIFAR-10 | accuracy (%): 85.74 -> 89.42 | Improved |
| add local learning to OPZO | OPZO · DVS-CIFAR10 | accuracy (%): 72.77 -> 77.47 | Improved |
| add local learning to DFA | DFA · DVS-Gesture | accuracy (%): 91.67 -> 91.43 | No clear effect |
| add intermediate global learning to OPZO (w/ LL) (ablation) | OPZO 9-layer conv · DVS-Gesture | accuracy (%): 95.83 -> 96.88 | Improved |
| add intermediate global learning to OPZO (w/ LL) (ablation) | OPZO 9-layer conv · CIFAR-100 | accuracy (%): 65.87 -> 66.13 | No clear effect |
| use Rademacher instead of Gaussian noise (ablation) | OPZO · CIFAR-10 | accuracy (%): 85.73 -> 85.69 | No clear effect |
| perturb before neuron instead of after (ablation) | OPZO · CIFAR-10 | accuracy (%): 85.73 -> 84.37 | Hurt |
| use OPZO instead of ZO_sp | OTTT online training · CIFAR-100 | accuracy (%): 22.26 -> 60.93 | Improved |
| fine-tune with OPZO vs direct test under noise 0.15 | NF-ResNet-34 · ImageNet | accuracy (%): 54.01 -> 60.96 | Improved |
| fine-tune with DFA under noise 0.15 | NF-ResNet-34 · ImageNet | accuracy (%): 54.01 -> 54.59 | No clear effect |
Closest papers by meaning
- Zeroth-Order Forward-Only SNN Training Inspiring Neuromorphic On-Chip Learning (ICML 2026 · similarity 0.91)
- NDOT: Neuronal Dynamics-based Online Training for Spiking Neural Networks (ICML 2024 · similarity 0.91)
- Towards Memory- and Time-Efficient Backpropagation for Training Spiking Neural Networks (ICCV 2023 · similarity 0.90)
- Spike-driven Transformer V2: Meta Spiking Neural Network Architecture Inspiring the Design of Next-generation Neuromorphic Chips (ICLR 2024 · similarity 0.89)
- Enhanced Self-Distillation Framework for Efficient Spiking Neural Network Training (NeurIPS 2025 · similarity 0.88)
- Multiplication-Free Parallelizable Spiking Neurons with Efficient Spatio-Temporal Dynamics (NeurIPS 2025 · similarity 0.88)
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.
