Research map / Adversarial robustness and security / Attacks on text and graphs
Rate Gradient Approximation Attack Threats Deep Spiking Neural Networks
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 rate-gradient approximation instead of STBP gradients for FGSM | VGG-11 SNN · CIFAR-10, direct input, T=8, λ=1.0, white-box | attack success rate: 86.2777 -> 93.7352 | Helped |
| Use rate-gradient approximation instead of STBP gradients for PGD | VGG-11 SNN · CIFAR-10, direct input, T=8, λ=1.0, white-box | attack success rate: 99.4949 -> 99.8281 | Helped |
| Use rate-gradient approximation instead of STBP gradients for FGSM | VGG-11 SNN · CIFAR-10, Poisson input, T=16, λ=1.0, white-box | attack success rate: 54.9798 -> 58.0328 | Helped |
| Use rate-gradient approximation instead of STBP gradients for FGSM | ResNet-17 SNN · CIFAR-10, direct input, T=8, λ=0.9, black-box | attack success rate: 67.1109 -> 80.1053 | Helped |
| Use rate-gradient approximation instead of STBP gradients for FGSM | VGG-11 SNN · CIFAR-100, direct input, T=8, λ=0.9, white-box | attack success rate: 92.8766 -> 94.7189 | Helped |
| Use rate-gradient approximation instead of STBP gradients for PGD | ResNet-17 SNN · CIFAR-100, direct input, T=8, λ=0.9, black-box | attack success rate: 87.6336 -> 95.2949 | Helped |
| Use rate-gradient approximation instead of STBP gradients for FGSM | VGG-11 SNN · CIFAR10-DVS, frame input, T=10, λ=0.9, white-box | attack success rate: 59.5084 -> 59.5607 | No clear effect |
| Shuffle each neuron's output spike order (ablation) | STBP-trained VGG-11 SNN · CIFAR-10, λ=1.0, soft reset | clean accuracy: 92.75 -> 92.086 | No clear effect |
| Shuffle each neuron's output spike order (ablation) | ANN-SNN-converted VGG-11 SNN · CIFAR-10, λ=1.0, soft reset | clean accuracy: 93.25 -> 93.358 | No clear effect |
| Extend attack-time simulation while keeping evaluation time unchanged (ablation) | RGA-based attack · CIFAR-10, VGG-11 SNN, especially Poisson-input setting | attack success rate | Helped |
| Extend attack-time simulation for an STBP-based attack (ablation) | STBP-based attack · CIFAR-10, VGG-11 SNN | attack success rate | No clear effect |
Closest papers by meaning
- Towards Effective and Sparse Adversarial Attack on Spiking Neural Networks via Breaking Invisible Surrogate Gradients (CVPR 2025 · similarity 0.89)
- Enhancing Adversarial Robustness in SNNs with Sparse Gradients (ICML 2024 · similarity 0.88)
- Advancing Training Efficiency of Deep Spiking Neural Networks through Rate-based Backpropagation (NeurIPS 2024 · similarity 0.87)
- S$^2$NN: Sub-bit Spiking Neural Networks (NeurIPS 2025 · similarity 0.87)
- EnOF-SNN: Training Accurate Spiking Neural Networks via Enhancing the Output Feature (NeurIPS 2024 · similarity 0.87)
- Robust Stable Spiking Neural Networks (ICML 2024 · similarity 0.87)
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.