Research map / Architectures and efficiency / Neural architecture search
Neural Architecture Search for Spiking Neural Networks
Youngeun Kim, Yuhang Li, Hyoungseob Park, Yeshwanth Venkatesha, Priyadarshini Panda
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 |
|---|---|---|---|
| add backward (temporal feedback) connections to searched cell (ablation) | SNASNet · CIFAR10 (T=5) | accuracy (%): 93.12 -> 93.73 | Improved |
| add backward connections to searched cell (ablation) | SNASNet · CIFAR100 (T=5) | accuracy (%): 70.06 -> 73.04 | Improved |
| add backward connections to searched cell (ablation) | SNASNet · TinyImageNet (T=5) | accuracy (%): 52.81 -> 54.60 | Improved |
| use Sparsity-Aware Hamming Distance instead of plain Hamming distance (ablation) | SNASNet-Fw · CIFAR100 | accuracy (%): 64.16 -> 70.06 | Improved |
| use Sparsity-Aware Hamming Distance instead of plain Hamming distance (ablation) | SNASNet-Bw · CIFAR100 | accuracy (%): 66.80 -> 73.04 | Improved |
| increase timesteps from 5 to 20 (ablation) | SNASNet-Fw · CIFAR100 | accuracy (%): 70.06 -> 70.52 | Improved |
| increase timesteps from 5 to 20 (ablation) | SNASNet-Bw · CIFAR100 | accuracy (%): 73.04 -> 74.24 | Improved |
| transfer architecture searched on CIFAR10 to TinyImageNet (backward) (ablation) | SNASNet-Bw · TinyImageNet | accuracy (%): -0.80 vs original search | Hurt |
| transfer backward architecture searched on CIFAR100 to TinyImageNet (ablation) | SNASNet-Bw · TinyImageNet | accuracy (%): +0.59 vs original search | Improved |
| transfer forward architectures searched on another dataset (ablation) | SNASNet-Fw · CIFAR10, CIFAR100, TinyImageNet | accuracy (%): -0.56 to +0.11 | No clear effect |
| use more than 2 backward skip connections (ablation) | SNASNet-Bw · CIFAR100 | accuracy (%) | Hurt |
| use average pooling in cells (ablation) | SNASNet-Fw · CIFAR100 | accuracy (%) | Hurt |
Closest papers by meaning
- Differentiable hierarchical and surrogate gradient search for spiking neural networks (NeurIPS 2022 · similarity 0.92)
- AutoSNN: Towards Energy-Efficient Spiking Neural Networks (ICML 2022 · similarity 0.92)
- EnOF-SNN: Training Accurate Spiking Neural Networks via Enhancing the Output Feature (NeurIPS 2024 · similarity 0.89)
- Rethinking the Membrane Dynamics and Optimization Objectives of Spiking Neural Networks (NeurIPS 2024 · similarity 0.89)
- STAA-SNN: Spatial-Temporal Attention Aggregator for Spiking Neural Networks (CVPR 2025 · similarity 0.88)
- Temporal Flexibility in Spiking Neural Networks: Towards Generalization Across Time Steps and Deployment Friendliness (ICLR 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.
